#74 - Anritsu MG3670B signal generator repair

hello repair time again this time I have unread sue mg 3670 be signal generator it was hidden here for quite some time waiting for its turn and when I initially checked it when I got it it had no where F output but now it seems to have some other problem let's turn it on so it boots up with all self-tests passing and by the way the screen flickering appears on the recording only I see it as nice and steady no problem at all so it boots up and it does not respond to front panel controls at all and they did respond before so it seems like we have more than one problem and unfortunately I couldn't find any service manual for this thing so we will have to manage without so let's take it apart and see what can we do here is the back of the unit by the way this thing is quite heavy there is a sticker here cautioned here we wait 18 kilograms so we have a couple of fans some outputs burst trigger input GPIB this is the 10 megahertz reference output and link to the inputs the link can be removed to use an external reference here we have some link between input and output of local 1.5 gigahertz oscillator this is a sticker and reads Corp made in Japan serial number and options 0 3 and 25 option 3 I believe with high stability time base I'm not sure what option 25 is and this is a list of extensions it seems like this thing should have three modules installed so I the covers and removing the top cover did not help much there are more internal covers everything is formerly shielded and that's why this thing is so heavy removing the side cover did not help as well and the same thing on the other side so let's look at the bottom here is the bottom of the unit and this seems to be the main distribution board with some connectors for the modules and on this side as well and some power distribution as well this board is marked DC to DC regulator with a bunch of test points with voltages marked on the board very nice this is a separate board called filter board and there is another one here also called filter board this is some sort of RF module and there is a small PCB here called divider filter so maybe this is some sort of a divider so I already checked the voltages here and they are very close to the specified values no problem here you I managed to take the front panel I would not sure this was useful but let's have a look anyway so here we have sharp LCD made in Japan and this is the control panel and maybe you can see here there are two boards one on top of the other I removed some covers here and check this out we have very serious RF plumbing going on here for massive fully shielded modules and one more module here two boards are visible and a smaller shielded RF can on this side this must be the ten megahertz reference we've seen connectors on the back there is a smaller RF box here under this cover I believe we have the main computer this is GPIB I see configurations which is LEDs and batteries to retain some configuration I suppose and I did not open these covers yet too many cables going to the front panel hem the back panel these free boards are marked modulation unit filter unit and modulation unit as well and this is a power supply here is a strange indicator on the power supply I have no idea what this is here is the CPU world there are four batteries here for some reason I measured them the fine three volt no problem at all this is the main CPU Toshiba CMP 68 HC 0 0 0 T 12 megahertz the software is revision 1.8 in the date codes and the chips are from the year 1995 and 96 and let's have a look on the other side some more logic here this is a CDMA modulation board 1 so we have a shoot at RF can 5 Xilinx FPGA for Texas Instruments DSPs and a bunch of memory and there is all the flux around the memory chips as if they were replaced at some point and also a lot of modifications all over the board and on this side as well here and here and the wire here this board is marked CDMA modulation unit 2 and on this side it is marked filter unit and we have I believe more DSPs here and this chip is also gspn by Texas Instruments some other logic and bunch of software some configurations which is here probably a test point or something and this is the third board marked here PI over 4 dqpsk modulation unit here we have a couple of unreachable branded chips a lot of other logic some configurations which is here and some more software check this out it occurred to me to start this thing without the modulation boards and here we go we have a responding front panel my guess is that something is wrong with one with the modulation boards so that means CPU could not initialize it and kept trying forever or something like that so I think now it's time to check some basic functionality I started checking this is connected to the reference output on the back of the unit through 6 DB attenuator here and no problem at all the 10 megahertz reference is fine so let's disconnect this and connect the output here I have set ten megahertz as well 0 DBM the RF output is on and we see nothing at all and I checked on different frequencies and there is nothing and this is the low from the back of the unit I removed this link we have seen before and no problem at all we have 1.5 gigahertz they say it should be minus 13 DBM and we read minus 6 DBM here so it's a bit higher but I would think it's not a problem [Music] let's have a closer look inside I believe this little module is the main oscillator so called the egg or yig oscillator which can be tuned in a wide range of frequencies in case of this instrument the range should be about 2.25 gigahertz but the lowest frequency of this oscillator is certainly not the lowest frequency of this instrument which is 300 kilohertz perhaps the lowest frequency of this guy is several gigahertz and at some point here this high frequency should be down converted to the final frequency range of this generator which is from 300 kilohertz to 2.25 gigahertz so let's check this thing disconnect this rigid coax connect spectrum analyzer and have a look I really hope it works otherwise I don't like my chances these things are either not available at all or too expensive and this can easily mean that this instrument is beyond economical prepare look at this we have a tone here and the signal is quite strong we have 6 DB attenuator here and the signal is just above 0 DBM and we have 5.39 2 gigahertz which is the lowest frequency because I set the generator to 300 kilohertz now let's at the highest which is 20 to 15 megahertz and now we have a tone here which is 7.62 7 gigahertz and the difference is about 2.25 gigahertz very cool I was not sure what frequency range to expect but this looks fine to me I am quite happy that the Eagle Slater is alive so the oscillator seems to be alive I connected this rigid coax back and it goes into this module here so it seems that this module is to control the yuga so later and this must be the output from it so let's check it I connected the spectrum analyzer here and essentially I see the same tones but about 10 DB lower now we are looking at the high end of the frequency range I am not sure what to expect here most probably this is fine so we have signal at this point and unfortunately I don't understand yet what's going on here lots of connections go in every which way and I don't even have a block diagram so let's try stepping back from the output I managed to remove the side panel here it is after removing a few screws it slides right out so here is the output connector on the front and I see rigid coax running to this point and this must be the output attenuator so let's try checking here just before the attenuator I see absolutely nothing at this point so the attenuator is connected to this module and the only rigid cogs besides that is this there is one more flexible quarks here must be a lower frequency one so let's check here so at this point we have about minus 20 DBM and there is no attenuator now so this seems a bit low and this is at the high end of the frequency range and look what happens if we go to the low end I will set 300 kilohertz here and we are down to minus 36 DBM let's increase the frequency little let's say hundred megahertz and now we have about minus 30 and let's try somewhere in the middle like 1 gigahertz and here we have minus 14 I would expect approximately the same level across the whole frequency range plus minus a few DB or something like that and of course I checked my connections and tried a different cable and there is no difference so I think something is wrong and I'm not sure what to do next let me think about it here I removed the last two modules to try to get more understanding what's going on here this one is the final module with the output and this one is just before it and it's called the convertor filter here this must be the input to it because that rigid corks from the each controller I was connected here and this must be the output because it was connected to this module let's start from this final module I already removed the screws here we are so the input is here coming from the other side of the module then signal goes through this box and then through this box and finally into this box and now it seems to me that these two things are they step attenuators and this must be for reverse power protection or something like that this is the other side again I already removed the screws and we have an RF amp here this is the input this is the output going through to the other side we have seen before and if we have a closer look at the input here is some passive filter to amplification stages there are two small active devices and then signal goes through this chip I looked it up this is 15 DB digital generator in 1 DB steps and the bandwidth of this guy is 0 to 2 gigahertz so they are using it slightly out of spec stretching it to 2.25 gigahertz but my point is that there is no down conversion here at all the input should be already down converted and we already checked the input and we saw that five to seven gigahertz signal with low amplitude dropping towards the low frequencies so that was clearly wrong it seems like that down conversion is not working properly and we see some sort of feed through this is the converter module and again I already removed the screws and I also removed the screws from these two shields so we can have a better look these two are identical mixers biased II microwave Zulu's let's remove them so they don't fall out when we start turning this module around there you go so this is our RF input again from about five point four to about seven point six gigahertz here it says drive amp on the board so the signal is amplified and it goes out right here here this goes through this filter which says five point four to eight point six gigahertz and comes back which is this point and it goes into the low input for some reason of this mixer and this is the iaf output so this must be the final result and this port was connected to the output amplifier and we saw incorrect results here so now let's look at this it says on the board 4.2 gigahertz PLL so this input must be some sort of a reference so it goes here and this thing is 4.2 gigahertz oscillator apparently can be mechanically adjusted so the result goes right here gets amplified and there is a splitter here so half of the signal I would think goes this way gets amplified and closes the PLL loop and I think this pin is tuning because I see it goes out right here and is connected to this pin of the oscillator this pin so it must be tuning another part of this 4.2 gig signal goes this way gets amplified field and goes into the ello input of this mixer and here things get a little weird because apparently this RF input is used as output and this is output as input and I believe so because I looked up this amplifier chip there are many of them on the board this is an EC amplifier and this is the input this is the output so the signal must go this way through this mechanically adjusted filter and it goes out here goes through this filter which is not marked at all and comes back right here which is this point and it goes into the RF input of this mixer so apparently hello and RFS swapped and let's see what does it take to convert five point four to seven point six gigahertz signal into the final result which should be from zero to two point twenty five gig apparently we need to subtract approximately five point four gigahertz and here we are here it says five point four gig amp on the board so it seems to me that this hello four point two gigahertz hello is mixed with something else and becomes five point four gigahertz here so this input must be about 1.2 gigahertz so all this kind of makes sense except some liberties taken with swapping pins of the mixers especially this one but I assume that an ritu engineer know what they're doing so now we need to understand what's wrong here and now I put this converter module back in and I'm checking at this point I removed this link between the sport and the this external filter which goes into the final mixer and we should see 5.4 gigahertz alow here for the final mixer and there is absolutely nothing there and of course the final conversion is not going to work without it this is the next module I remove the screws here as well let's see here we have a modulation board and I think these two are modulation inputs and this is some sort of a heated chamber let's see if I can remove this there you go here we have one own power registers for sitting and let's see this must be the reference input and at this point it must be hundred megahertz because these two are hundred megahertz crystal filters and here is the reference output and this goes through the converter module and also the reference must participate in this modulation process somehow and I see several are connections to the other side like here and here so let's have a look I already removed the cover here this is 1.5 gigahertz generator this is the output this goes to the back panel and there is a jumper and it comes back here and this must be modulation and they combined here somehow and at this point we must have 1.2 gigahertz because these three are narrow filters marked 1200 and this must be the modulated 1.2 gigahertz signal which goes to one of the mixers in the converter module and let's see if I can remove this one as well there you go and the sensor is the same thing for power resistors to heat up some chamber this is the reference output from the modulation board that goes to the converter board so we see hundred megahertz and around seven DBM this looks excellent and this is the 1.2 gigahertz output from the modulation board the frequency is perfect but the amplitude looks a bit low it is about minus 20 DBM and this is one of the inputs to a mixer and I would expect something higher than this let's say something closer to 0 DBM here is the converter module again now let's summarize what do we know so far we know that the input signal here is fine the reference is fine and the output is bad essentially we see a weak signal from the input leaking through and there is no 5.4 gigahertz alow at this point we measured at this point and there is nothing at all here and this must be a result of mixing here going through this filter and this amplifier and we also know that this signal 1.2 gigahertz is present but the amplitude seems to be low so we need some way to start checking around here the problem might be because of this low input here or maybe this 4.2 gigahertz signal is missing let's say this oscillator might be not working or some amplifiers here not working or maybe this mixer is bad so I noticed this screw here covering a hole alright yeah so I'm thinking about passing pink walks through this hole and attaching here to the output from the mixer so we can check here by passing this filter and this amplifier let's try here we are I passed this coax for this hole attach the ground here attach the signal here and I know this is far from perfect because all the distances here let's say this tree is going nowhere and this one and this weird shape here they are all important and the distances between them and the ground and between each other are important and I introduced more ground here and extra load at this point certainly disturbs a matching but we need some visibility into the problem and I'm hoping that we will learn more than we know now and here is what we've got at that point we are looking at the frequency range from 0 to 6 gigahertz and there is nothing at all here except this tiny blip around 300 megahertz and about minus 57 DBM or so I don't see that 1.2 gigahertz a low feed through perhaps it is below the noise floor it was about minus 20 DBM to begin with but still I would imagine it should stick out a little bit and there is no 4.2 gigahertz feed through as well and no intermodulation products now I moved this test cable from this point to this point and before the mixer and even before this filter on the board to see what's going on here with the 4.3 gigahertz input to the mixer and to pass this cable here I removed this shield again from the mixer and attached the cover of the mixer with a couple of screws so let's put the cover back and have a look well we cannot arrange from three to five gigahertz and there is nothing here at all and now I removed this shield between sections here and passed this cable all the way to the output from the oscillator I'm so glad these removable dividers exist here so we can check what's going on here with closed cover let's go and check this out here we do have around minus 17 DBM and about 4.2 gigahertz signal 4.1 87 but the resolution now is so crude we are looking at the range from 3 to 5 gigahertz let me zoom in to have a better measurement here we are now we have hundred megahertz span and we can see that the frequency is almost exactly 4.2 but it's not stable it's shifting up and down a little bit I think something is wrong with PLL and now I set the span to 10 megahertz and we can clearly see that the frequency is shifting [Music] and now we are looking at the slater tuning pin here on the scope we have 10 volts per division so the top of this waveform is around 12 volts which is one of the power rails of the oscillator so it seems to me that the PLL is trying to pull the frequency and is hitting the power rail for some reason and the frequency is about eight and a half Hertz so something seems to be wrong with PLL now I remembered seeing this jumper here which I did not mention I did mention I believe that this pin seems to be the output from PLL that goes to that tuning pin of the oscillator but not directly it goes through this jumper and if we moved the jumper then the PLL was disconnected and the tuning pin is connected to some fixed voltage so this must be for testing let's try so I moved the jumper and sure enough the frequency looks stable now I set this pan to 1 megahertz and we don't see any movement and the frequency is four point one nine seven nine one five which is slightly more than two kilohertz off and I'm not sure if it's close enough or not maybe PLL is asked to pull too far and that's why this heating the rail if so then we need to adjust the frequency of the oscillator so we might have more than one problem here before worrying about the frequency I would like to find out why there is no signal at this point perhaps one of these amplifiers is dead so let's try to move this test cable let's say after the first amplifier and see if there is signal at that point I moved the cable to this point after the first amplifier and let's also check the power rails here I see that these two amplifiers receive power from the same power rail which is this pin so we should see this pin on the other side its marked Ln g8 and the power goes through this filter and then there are two inductors one for this amplifier and one for this one so now let's check what we have here look at this we have two very low amplitude tones at this point this one is about four point three three gigahertz and about minus 25 DBM or so and the other one is about 4.2 gigahertz and slightly lower so this one seems to be the oscillator and this one might be I guess an inter modulation product with the reference because it is about 100 megahertz higher which is the frequency of the reference and the frequency measurement now is not very precise at this resolution we have 500 megahertz punt so something seems to be very wrong here perhaps this amplifier is dead or something like that and the PLL might be fine or not but it seems like the signal does not even reach the PLL and now let's check the power here's our ground and this is the power pin and there was our problem point two volts here on the other side of the board is a regulator and these three pins are they regulate appeals this must be the input twelve and a half when this is the output point two volts so either this regulator is dead or the amplifier is overloading it so I just soldered this wire from the pin and here we have the voltage from the regulator which is a little bit too low here is another amplifier in the five point four gigahertz path so I'm still not sure what's wrong here regulator or the amplifier one way is to try loading this regulator but we need to know the current and another way is to inject a voltage into the amplifier from a lab supply but we also need to know a sensible current limit let's see if I can find a datasheet for this amplifier I found a datasheet this is a 2 to 8 gigahertz wideband amplifier about 15 DB gain and here we have a supply current typical is 135 millions and 180 is the maximum and seems like 8 volts is the typical supply voltage and now I'm providing the it will trail from this lab supply I set the current limit to 400 milliamps just in case and we can see that the actual current is 244 almost 245 milliamps or about 122 each amplifier which is even slightly less than the typical 135 according to the datasheet so there is no problem here no overload now let's check the RF signal yes the signal is nice and strong this is after the first amplifier around 0 DBM no problem at all now we need to fix the voltage regulator and now I'm checking this voltage regulator using this electronic load and it puts out 8 volts let's try to load it with hundred milliamps for now and there is no problem at all let's increase 200 300 no problem I'm not sure what was wrong with it before let's have a look here is this voltage regulator perhaps you can see it and I let it run under the 300 million load for quite some time and there was no problem at all so I'm starting to suspect maybe you can see some sort of contamination on the board all around this place and there is ground plane around this point of connection of this little wire so maybe there was some short somewhere and I somehow removed it during soldering I'm not sure what's going on here I am tempted to clean the board all over this area and solder the wire back and try again so I clean the board soldered that wire back to the pin and there is no problem with the power rail now so I moved that PLL jumper into the working position and we have nice and stable signal now 4.2 gigahertz exactly with 500 kilohertz span and the tuning voltage is also nice and stable sitting around 4 volts this looks excellent I just found that this 1.2 gigahertz signal at this port which we looked at before and it was low about -21 DBM or so I found that it was low because I did not connect these two modulation input signals and it turns out this is important for some sort of mixing process here in this module when this 100 megahertz the reference signal is tripled and mixed with the 1.5 gigahertz alow and the result is this 1.2 gigahertz output with some sort of modulation writing on top of it and even if we are not really modulating anything I guess there are some carriers present anyway and they are important to have in this mixing process here is what we have at that port with the modulation signals properly connected about minus 3 DBM and with some loss in this one and not very good cable so this looks much better and now we are back to the converter module I removed the test cable and now we are looking at this port which is five point four gigahertz a low output just before it should go through that external filter which we don't see because of the frame into the final mixer and here's what we got nice and strong minus 10 DBM signal and I'm not entirely sure what level we should expect but this does not look bad at all so let's put this lid back in place and check the result of the final down conversion this is the output from the converter module and the frequency is set to the maximum which is 2.25 gigahertz and that the frequency is not quite right 2.2 to 8 8 something and this analyzer has a very good reference so this is suspicious perhaps some adjustment is needed and the level is about minus 24 DBM and I'm not sure if this is correct or not so let's try with the output amplifier module in place so now with the output amplifier in place we have the right output level here on the front panel output we have 0 DBM set here and we are reading minus 1 DBM here and I can easily imagine 1 DB of loss in this cable and two adapters so this is great and we can try changing the level here let's say - -10 - 11 here let's go to - 20 - 21 - 30 - 31 perfect let's go back to 0 so the level is right but the frequency is quite a bit off I start exactly 2 gigahertz here but here we have 1 9 7 8 7 5 which is quite a bit off and this analyzer has a very good reference so this is incorrect and look what happens if I change this frequency just by 1 Hertz down it jumped up and quite a bit we have a span of 50 megahertz here so it jumped by probably tens of megahertz and if I keep going down and does not jump like that anymore but if I set 2 gigahertz again it jumps down again so something is not right with the frequency control perhaps with the niggas'll later control module here we have two more modules we never looked at before let's start with this one it is marked RF quant one on this side and Aref quant two on this side there are two words and this probably means RF controller or something like that this connector is for eager salita coils this is input from Higa Selita and this is output to the converter module this goes to this module and here is a cable here that goes back and connects here and let's see this one is marked local board one here and local board two on this side also two boards almost fully shielded this is the input from a reference and this goes out to the reference input of the converter module so let's see we can understand what's going on here so I remove this cover here and there is one more cover and it was stuck and when I managed to carefully lift it here is what I found this RF absorber somehow deteriorated and the piece of it is stuck to the board next to the amplifier chip and I wonder how did this happen could it deteriorate by itself or this as a sign of overheating anyway let's have a look what's going on here so this is the input from ego Selita then I see a splitter here and one part of the signal goes out and this goes to the converter module and another part goes to the amplifier this is called isolation amp here on the board so the signal is amplified filtered here and goes to the other side here is the other side I already removed the covers this board is called sampler and at this point that signal from the isolation amp comes in it goes here and this looks like a mixer so this must be another input to the mixer coming from the other module called the local board the result goes here this is an amplifier and this Motorola chip is a phase detector specifically made for precision phase locked loops there is some adjustment here this was covered as well with this little cover this is a 12 bit burr brown duck and these four are identical hundreds of branded chips I'm not sure what they do I found the service manual for mg36 9x line of signal generators they can go up to 20 gigahertz so this block diagram looks quite different of course but I believe this part looks very similar and this is eager solely to control so it can be very helpful to understand better what's going on here is the description and the block diagram of that part so let's rotate it and have a look there are several phase locked loops here one loop is this reference loop to produce hundred megahertz from ten megahertz reference this is ego so later and I believe this part looks a bit different in our case we have seen that the insulator output comes in gets split right away and one part goes right out to the Duncan Burton module and the other part goes through the isolation amplifier and into the sampler and the other input into the sampler is from the scores loop and it goes through this step recovery diode here to produce a lot of harmonics so the idea here is to program from the main CPU this divider in the course loop so that the resulting frequency will have one of the harmonics quite close to the desired frequency so after mixing with this English Slater signal here the result would be in this range from 21.5 to 40 megahertz so this result goes into this phase detector and it is compared with this signal which comes from this fine tuning loop which is also programmed from the main CPU and to produce fine tune frequency which is compared here and the difference should drive the fine tuned coil in the IGAs later and these frequencies in our keys can be a bit different but the overall idea should be the same so this input into the sampler must be from the course loop and this input must be from the fine loop and here we should see the output from the center I have this BNC adapter which should fit these connectors and we can solder a cable at this point with check what's going on here this is the input to the sampler about 97 megahertz and a lot of harmonics this looks quite good and this must be the fine union signal about 12.5 megahertz and also with a lot of harmonics and I'm not sure harmonics here are intentional but I think this looks good as well and this is the output from the mixer in the sampler the amplitude is low - 36 DBM or so and the frequency is different about 13 megahertz and I connected the signals back so the phase lock loop should be locked but apparently it is not so something is wrong here I took a closer look at this board and in particular under this little color and I think I see something that I didn't notice before this capacitor which is mounted horizontally for some reason I believe has some corrosion on the legs and it seems to me there is some residue on the board around it let's check this capacitor with LCR meter capacitance mode one kilohertz let's enable ESR as the second parameter this is 10 microfarad 35 volt capacitor 1 point 8 micro farad 40 ohm ESR this looks quite bad so let's do solder it and check out of circuit sometimes it can be misleading in circuit it looks even worse out of circuit 700 nano farad and about 78 ohm SR and it might be not just because it is out of circuit but because it was heated up during the soldering and more electrolyte evaporated or something like that and it doesn't matter clearly this capacitor is dead I installed a new capacitor and look at this this is the output from the error of control module which should go to the down converter module I didn't want to install that and connect all the plumbing just yet and we have exactly 7 point 4 gigahertz here as expected given 2 gigahertz here and now let's see what happens if I adjust this by 1 Hertz down as before and back and down again and back you see this frequency jumps briefly and comes right back and I think this is because the course loop frequency is reprogrammed at this point and the fine loop is adjusted accordingly of course and the frequency is unstable for a brief moment until the PLL looks again and this is normal given the circuit texture with coarse and fine loops we must have multiple such points alone the frequency range of this generator and now let's try changing this frequency now let's see we have ten megahertz punt here so one megahertz per division so let's go down by one megahertz nine nine nine there you go it jumped exactly one division let's enable the counter to get exact frequency seven point three nine nine exactly so this looks perfect and now with the downconverter module back in place and the output from the front panel maximum frequency 2.25 gigahertz the level is correct the frequency is correct as well and let's try changing the frequency by one megahertz which is one division point two four nine gigahertz and yes it jumped by one division exactly let's see two point two four nine gigahertz looks great so everything is back together here but I didn't worry about the options yet this video is getting too long maybe I'll make another video troubleshooting the options if you like this one and give it thumbs up thank you very much bye

Channel: FeedbackLoop

Views: 11,466

Rating: 4.9816513 out of 5

Keywords: diy electronics, electronics repair, PLL, phase-lock loop, yig oscillatior

Id: v8wE2MqeiXM

Length: 61min 30sec (3690 seconds)

Published: Sun Jul 28 2019