-
GW Instek GOS-6200 200MHz Oscilloscope - repair log [FIXED]
Posted by sean0118 on 04 Oct, 2021 03:07 -
The scope repair marathon continues.
I thought I'd crack open this GW GOS-6200 because I'm waiting on parts to fix my Tek TDS2024, Tek 7603 and GW GDS-2204.
I haven't powered it up but I was told the flyback had failed, but I don't think anyone has attempted to fault find etc. So I assume it just means the CRT doesn't power on and the flyback may or may not be okay.
Turns out it has a similar issue to the GDS-2204, the epoxy used on the power supply board has gone conductive and corrosive. I ended up pulling all of the caps to clean it up. I tested them with some smart tweezers, they seem pretty much okay, so I don't think they have leaked, all of the corrosion is from the epoxy. They are Jamicon SK series which aren't great, but not too bad. Strangely C1023 (1uF, 160V) is a TK series part even though there is a 1uF, 160V SK right next to it. The TK part seems to be double the ESR, 30R from memory but I might be wrong.
Some of the traces under the rectifiers looked a bit crispy, although the rectifiers test okay, I need to investigate this further.
The glue had got in under one of the rectifiers and caused corrosion to the pads, so I pulled it to clean it up properly.
Power Supply Board:
Unfortunately they also used epoxy to glue on the connections to the vertical and horizontal deflection plates. Looks like the corrosion isn't too bad, might have got to it just in time. The lug that grounds the CRT was also a bit loose, I'll add a lock washer to this.
Vert and horizontal deflection:
The main board looked okay, it also uses Jamicon caps, although I probably won;t replace these. They've used the glue here again to attach a coax cable, so I scraped off as much as I could and added DeOxit.
Main Board:
So turns out the CRT is a D14383GH-123, looks like it was also used in Fluke scopes.
The HV board looks okay, although would be difficult to tell if there's a fault inside the flyback module. The switching transistor is a UTC 2SD880 NPN, tested okay just using the multimeters diode mode.
There were only three Jamicon caps in the HV module so I replaced them with Panasonic and Rubycon parts, although the Jamicon parts did test okay.
CRT & HV Board:
The next step is to test some more components on the power supply board. The crispy rectifier traces might indicate an issue, hard to say. I've already pulled all of the electrolytics from the power supply board, so may as well replace them with better quality parts too.
-
Ok, time for an update.
I recapped the whole power supply and the vertical amp board (except for one non-polarised cap).
I powered it up and all of the rails on the power supply look good, very low ripple too (+110V, +55V, +12V, -12V, +5V). Except I'm not getting a proper trace, it appears to be scanning off the bottom of the screen (I can see the glow of it sweeping if the intensity is turned right up). When switched to X-Y mode and set to high intensity I get an out of focus vertical line in the centre of the screen.
Because it's sweeping I assumed the horizontal amplifier was okay, so I probed the outputs of the vertical amp (Y+ measured at R1159 and Y- at R1156). Y- (blue) isn't being driven at all and is sitting at 55V (the supply), but Y+ (yellow) is being driven.
Is it Y+ which has an issue? If it's set to X-Y mode with the inputs set to GND coupling should both Y+ and Y- both be at 55V and no signal? There shouldn't be an offset between them right? Or should Y- be driven the same as Y+?
I have the schematic (attached to previous post) so should be easy to troubleshoot, I just want to check I'm on the right track first.
Vertical amp outputs:
edit: I also tried all of the focus and astig pots, but with very little change -
Okay, another update, turns out the transistors (2SC1252) which drive Y- are dead.
-
I put in some 2N4238's temporarily and it came to life!
The 2N4238's are too slow, but I'll have a look and try to find a proper replacement soon.
-
Okay, another update, turns out the transistors (2SC1252) which drive Y- are dead.
Found BF259 to be a good GP plate output amp. Check their specs against what you need. -
Okay I found some 2N3866's in a junk box at work which seem to be good replacements for the vertical amp.
The next issue I'm looking at is the horizontal scaling at divisions around 1ms. The horizontal amplifier board only has a "5ns ADJ" trimming capacitor, I assume this is for some high frequency timing or compensation.
All the other horizontal trimpots are under the bottom cover on the main board. I thought this would be an easy fix, but I've reached the end stops on both the "H_GAIN" and "1ms/div_ADJ" trimpots and the scaling is still out. Is this a common issue? Have the trimpots or other resistors drifted? Or would there be some reference frequency I need to adjust somewhere?
1kHz from HP 226A Time Mark Generator:
1kHz ref from Rigol DS1074Z:
Trim pot locations:
-
Okay, I've been probing around the adjustment trimpots and I found the horizontal ramp waveform. If the GW is set to 1ms, shouldn't the rising part of the ramp be for 10ms because the GW has 10 horizontal divisions?
Although, I would have thought only having 9ms ramp would give me the opposite problem to the scaling issue I have. Maybe it's the amplitude I should be focusing on?
-
Great work and looks like you caught that right in time. I've seen that glue eat the legs off components and eat away the tracks on the circuit board.
Looks like this is coming along well and looks to be a very useful scope. -
Okay, I've had another look at this, but haven't made much progress.
I realised the trace width only overscans the division lines by a little bit, not enough to explain the horizontal scaling issue. I suspect this means the period of the ramp waveform is roughly correct, but I'm only guessing...
I probed the ramp signal coming in and out of the "1ms/div adj" trimpot. The trimpot (1k) and some series resistors (120R, 2.7K) reduce the amplitude of the ramp. I pulled one leg of the 2.7k which let me measure their resistance, all three seem okay.
The ramp then goes into some transistor based inverter circuit (is inverter the correct name? It's level shifted to +5V and the ramp slopes down). I guess it's possible the ramp on the output of this inverter isn't being pulled up enough, which is what I'll look at next.
I might re-cap around that area of the PCB too. But it is tempting to just replace the 1k timmer with 2k or similar, but that feels like cheating...
Let me know if you have any ideas, I'm not really sure what these waveforms should be looking like...
Trace width:
Horizontal ramp going into trimpot (1ms/DIV):
Output of trimpot (1ms/DIV):
Output of inverter (1ms/DIV):
-
Typically setting up the timebase on a CRO requires an adjustable mark space generator so to get pulses to line up with graticules. Today any reasonable AWG easily can do this.
-
So I have a time mark generator that I borrowed, it agrees with my newer Rigol so I assume it's correct. The adjustment trimmers in the GW seem to work, but they reach the end stops before it's calibrated correctly.
I thought about it a bit more and I realised both the positive and negative outputs going to the horizontal amp board are roughly the same amplitude. So I think if the amplitude of the ramp is wrong it must be before the inverter circuit I mentioned earlier.
One interesting thing I did notice is it looks like there's heaps of high frequency noise on the inputs to the horizontal amp board. At first I thought it was pickup from measuring incorrectly, but even when using short gnd probes it stays, so I think it is really there and not measurement error:
-
So I have a time mark generator that I borrowed, it agrees with my newer Rigol so I assume it's correct.
Oh, so you do, sorry.
Yeah that looks gawd noisy however if you look at it faster does it align with any of the supply frequencies as it might be just the scope picking it up and not part of the sweep ramp at all ?
Looking back at previous screenshots and particularly this one:
Can you not adjust a whole period between the graticules as there should be like in the attached ?
-
Yeah that looks gawd noisy however if you look at it faster does it align with any of the supply frequencies as it might be just the scope picking it up and not part of the sweep ramp at all ?
Yeah I think I'll need to investigate it a little bit more, might be from the PSU as you say. The vertical amp has similar noise from memory.Can you not adjust a whole period between the graticules as there should be like in the attached ?
Almost but not quite, that's as close as it gets before the trimmer hits the stops. Pretty tempting to increase the resistor in series with it.
I actually forgot I had the schematics for this (attached to first post). I went back and had a look and noticed something interesting, what do they mean by "0.65V/div" in this context? SWP_OUT is the horizontal ramp signal where I'm getting close to 10Vpp.
-
There should be a SWP_OUT waveform somewhere in the documentation and 0.65V/div refers to the monitoring scopes vertical settings to replicate the same waveform.Can you not adjust a whole period between the graticules as there should be like in the attached ?
Almost but not quite, that's as close as it gets before the trimmer hits the stops. Pretty tempting to increase the resistor in series with it.
I actually forgot I had the schematics for this (attached to first post). I went back and had a look and noticed something interesting, what do they mean by "0.65V/div" in this context? SWP_OUT is the horizontal ramp signal where I'm getting close to 10Vpp.
Unusual yes that 0.65V/div is used however until you line up the example waveform with what you actually get it might not be clear why this is done.
How much additional resistance do you envisage needing in series with the trimmer ?
Could it just be things has drifted some and their totals marginally exceed spec ?
-
There should be a SWP_OUT waveform somewhere in the documentation and 0.65V/div refers to the monitoring scopes vertical settings to replicate the same waveform.
Unusual yes that 0.65V/div is used however until you line up the example waveform with what you actually get it might not be clear why this is done.
Thanks, that makes sense, I don't have the service manual unfortunately so I'm not sure what to set the monitoring scope too then.How much additional resistance do you envisage needing in series with the trimmer ?
Could it just be things has drifted some and their totals marginally exceed spec ?
I guess it probably isn't that much as the trimmer is only 1k, so maybe a few more 100Ohms? I might give it a go and report back.
I've also been recapping that area of the board, but surprisingly difficult. The pins of the 47uF are bent flat against the board, I've damaged three pads already even after trying to straighten the leads out, they just pull the pads with them. I have a repair kit though.
-
It could be the trimmers need some exercise too and some also give them a drink of IPA or Deoxit.How much additional resistance do you envisage needing in series with the trimmer ?
Could it just be things has drifted some and their totals marginally exceed spec ?
I guess it probably isn't that much as the trimmer is only 1k, so maybe a few more 100Ohms? I might give it a go and report back.QuoteI've also been recapping that area of the board, but surprisingly difficult. The pins of the 47uF are bent flat against the board, I've damaged three pads already even after trying to straighten the leads out, they just pull the pads with them. I have a repair kit though.
Yeah that's a pain when legs are hard bent over generally for wave soldering and probably with that Pb free muck in which case you'd be well advised to flood the joint with some leaded solder first to lower the melt point to help save ripping pads up. -
Yeah that's a pain when legs are hard bent over generally for wave soldering and probably with that Pb free muck in which case you'd be well advised to flood the joint with some leaded solder first to lower the melt point to help save ripping pads up.
The 47uF seem particularly bad for some reason, not sure why, all the others came out okay.
I replaced the 2.7k in series with the 1ms trimmer with a 3.3k. I can now trim it at 1ms, but the problem is the trace stops at 9 divisions and doesn't reach the edge of the screen. Then if I switch the time make generator and scope to 1us it's out again.
I'll need to think about this a bit more, I thought it would work but didn't really have the desired effect.
-
H Gain would seem like it needs adjusting to at least have a blank sweep extend past the graticules.
It's between Q905 and 906 in the Hori Preamp schematic on P26. -
I had a bit of a look at this again recently, not sure when I'll get back to it so here's some notes....
I found that the ramp used for the horizontal sweep was being reset too early, which happens when the base of Q618 reaches 0.6V. I had a bit of a look around this reset circuit and found R656 in my scope was actually 4.53k, not 4.02k as listed on the schematic.
The ramp is triggered by flip flop U601, which goes high until the reset circuit triggers. Which seems to be working.
I suspect the issue is the slop of the ramp is steeper than it should be, so it hits the limit and resets too early.
Unfortunately part of the ramp gen circuit was cut off when the document was scanned. But afaik the amplitude of the ramp depends on the current source and current sink (U607). Resistors are switched in to set the current in the sink for different horizontal sweep rates.
It should be pretty easy to calculate what the currents should be for different sweep rates, so I'll do that next. I might also pull the integrating cap C608 to test...
-
Hi!
Unfortunately all of these schematics are missing an inch or two of the middle bits, presumably because they were drawn larger than standard sizes of paper, and whoever scanned it didn't get it all on his scanner and never checked it was complete afterwards!
The only way to get the missing information would be to trace it out from the instrument itself, and alas to say the last one for sale was far more than I could afford to buy to draw out!
Chris Williams -
Okay, time for an update.
I think I've found the cause of the wrong horizontal sweep rates... the integrating cap C608 has drifted from 1 uF to 0.75 uF !! I should have pulled it to test much earlier, but thought being a film cap it'd be unlikely to drift much...
I found it because the current being drawn by the current sink (R655A // R636 // R637) was around 0.71 mA. I couldn't find any issues with this part of the circuit and even traced it back to the ADC and checked its 2.5 V ref.
But looking at the ramp signal on C608 I was measuring a voltage rise of 8.32 V (+0.56 -> +8.88 V) over 8.68 ms. That suggests a constant current of 0.96 mA if C608 is 1 uF, so doesn't match the 0.71 mA in the current sink. However, with C608 being only 0.75 uF it's 0.72 mA, so very close to the current sink. Mystery solved I think... bad C608...
I've attached an updated schematic with some of the missing circuit filled in. I've also attached photos of the front panel control board (which is where the ADC which drives the current source etc is).
-
I replaced C608 with a new 1uF 1% polypropylene capacitor (Vishay MKP1839510161 ) but.... it's still not fixed
I calibrated the horizontal amp and it now works very well down to 0.2ms, but switching to 0.1ms and below causes a timebase error again...
What seems to happen is at 0.1ms C608 is disconnected by switch U606 and a smaller integration cap is switched in. It's a pair made up of C603 and C605 (schematic cuts it off, but I think they're 18pF + 180pF). Same current sink is used for both. Q605, Q606 and Q607 seem to be an amplifier (I think?) which drives SWP_OUT as before.
I can't see any way to cal out a difference between the low and high frequency integration caps? Another odd thing is C605 is a silver mica, but C603 in parallel is just a ceramic, why not use a 200pF silver mica? Maybe the idea is to substitute different values for C603 if required?
What's confused me even more is when probing with another scope the 1ms horizontal ramp is x10 slower than the 0.1ms ramp, which seems right to me, so not sure why there's a big error in the display....
My current plan is to substitute larger values for C603, or maybe replace with a trimmer cap. Any other ideas?
-
I saved some screenshots for the 1ms/DIV and 0.1ms/DIV horizontal ramps (SWP_Out measured at R655)...
The ramp time is x10 quicker for 0.1ms/DIV which seems right. But the amplitudes are different... which means the slope isn't x10 (0.698 V/ms vs 6.77 V/ms)
What confuses me is the ramp is reset by the circuit around Q618, which is after the switch U606. That part of the circuit should be the same for both 1ms/DIV and 0.1ms/DIV... I must be misunderstanding something
Edit: Pretty sure there's about 5pF too much capacitance in C603 / C605. I haven't pulled C603 / C605 to check, but I think there's about 1.4uA running through the current sink, so must be about ~205pF to give that ramp. I think I want closer to 199.25pF... if I replace them with a 200pF cap I should get a slope closer to 6.98 V/ms...
1ms/DIV ramp: delta V = 8.12V, delta t = 11.64ms
0.1ms/DIV ramp: delta V = 7.79V, delta t = 1.164ms
-
what is an awg?
-
Arbitrary waveform generator