HP4194a bridge unbalanced repair

[ArjenCNX]

Ladies and gentlemen, and none of the above,

Currently im in the progress of repairing a bridge unbalanced fault on a vintage impedance analyzer, it seems a tough one to crack, after spending about 10 hours walking thru the schematics, i want to throw this one in the hive-mind, hoping some kind grey-beard will help me out.

Current progress:

- PSU checked, all good
- Logic boards function well
- Signal generation IF, etc seems within specification

Error occurs when i connect Lcurr and Lpot together, at low frequency the singal over Lcurr becomes corrupted, and a bridge unbalanced fault appears, together with some strange peaks on the readout.

The signal on Lcurr is about 6v p-p and clean, low distortion.  once i go to 100HZ and have the probe connected to Lcurr+Lpot, the signal distorts.   This happens inside board A23.   
I have checked the sense resistors and the switch boards but they seem good so long A23 is disconnected.  This makes me to believe the fault must be hiding in A23 somewhere.   ive checked all parts (in-circuit) so far and have not found any obviously wrong.  all caps seem OK,

I have bought the schematics online, so i have a readable document to work with.    the most obvious candidate for a problem is a transistor i think, but i cannot seem to find any that are obviously malfunctioning.

The hope here is that there is a grey beard in the team that has wrestled this beast, won and is happy to share his victory.

 

[ArjenCNX]

i have also found that the =12 and -12V rails are at 10.8V  but the LDO's have 12.1V output, indicating high current draw from somewhere,  IR camera shows that Q19+Q21 are at about 65 degrees

[ogden]

The signal on Lcurr is about 6v p-p and clean, low distortion.  once i go to 100HZ and have the probe connected to Lcurr+Lpot, the signal distorts.   This happens inside board A23.

It does so only at low frequencies? If high freq operation is OK, then electrolytic capacitors in the signal/IF/LO chain can be faulty/off_spec.

[ArjenCNX]

Thank you for that nugget Ogden, I’ll check some more capacitors, I’ve checked a good amount of caps already and I have measured them with a lcr meter finding they are all still good thus far. This is a beast of a meter, 37KG of electronics and it’s complex to wade thrue the 436 page service manual.  I do think next on the hop is to replace some of those hot transistors, I think 68 degrees for a TO92 is a bit much and the power supply there is obviously beeing drawn down since the LDO’s are at their voltage but down the line it loses 1.2 volts or so. The Ir camera does a good job In capturing that.  If there are any other leads I’d gladly hear them this must be the most complex device I’ve fixed thus far next in line would be my dynamic signal analyzer, it’s tantalums have reached their end by now and it has hundreds of them. 

[ogden]

I think 68 degrees for a TO92 is a bit much and the power supply there is obviously beeing drawn down since the LDO’s are at their voltage but down the line it loses 1.2 volts or so. The Ir camera does a good job In capturing that.

IR camera is nice to have indeed Sure check hot transistors - why not. As you suspect A23 board, then good idea is to do level measurements described in Table A23-1 (page 8-158). Check other levels as well, especially LO's. For example TP5 on A23. If C32 and C42 is dried-up then LO level could be degraded - in result IF level will be lower and so on. Other important electrolytic capacitors on that board are C27, C31, C14, C5.

[Tomorokoshi]

Error occurs when i connect Lcurr and Lpot together, at low frequency the singal over Lcurr becomes corrupted, and a bridge unbalanced fault appears, together with some strange peaks on the readout.

The signal on Lcurr is about 6v p-p and clean, low distortion.  once i go to 100HZ and have the probe connected to Lcurr+Lpot, the signal distorts.   This happens inside board A23.

I'm not quite clear on what your setup is here. Can you add pictures with the setup in the distortion case, exactly how the probe and scope are connected, etc.?

What is the 6v p-p measured relative to, and where are you getting that signal? From A23 or from Lcur on the front BNC connector?

[ArjenCNX]

Thank you for your comments,   well i've dug a bit deeper and found that the bridge unbalance circuit is fed with a nice sine-wave from A22, this sine wave distorts as soon as it passes Q18/19/20/21 (lower right corner in schematic) also Q18+Q20 are the ones that heat up, im quite sure they need to be replaced as its just them plus the resistors in series with them that get's hot.  also other similar transistors in the device do not run hot, even though they have the same part number.  current flow is 56mA in the transistor and it runs up to 70 degrees, im sure HP engineers would not use the parts at there upper limits.   now the next puzzle piece needs to be found,  what transistor is equivalent to the HP# 1854-1073 ?  i've checked several large cross reference databases but came up with nothing.  its not a regular transistor with a  bandwidth of 6.5GHZ !!   Question is weather i need that much bandwidth, since the highest frequency inside this device is 160MHZ, its probably the linear characteristics they where after when selecting this part.   

OK just figgured it out,    2SC3355  ..........     lets see if i can order those puppies!

[ArjenCNX]

Replaced the transistors, but no luck, the problem persists,  seems to be getting really complex now since the problem is hard to define.  it seems an issue only at frequencies below 500HZ, but i have no good reference here that i can compare it with.

[ogden]

Replaced the transistors, but no luck, the problem persists,  seems to be getting really complex now since the problem is hard to define.  it seems an issue only at frequencies below 500HZ, but i have no good reference here that i can compare it with.

What about electrolytic capacitors (I mentioned) in the signal/IF/LO chains? Signal electrolytic caps that lost their capacitance are #1 reason in case of degraded low freq performance. It may help if you pick low frequency that does not cause unbalance error, like 1000Hz and measure signal levels in all relevant testpoints, compare with service guide.