HP3310 repair: Output power amp and feedback path

[coldfiremc]

Hi. I've recently received a Hp3310A Function generator. The detail was: output connector broken. Mostly worked ok, just a little frequency shift, nothing really important.

So I've bought one of those bnc chassis connectors, opened and to take out connector I had to disconnect the "power" output transistors (chassis mounted). Transistors were full of rust, so I tried to take out one of the pins from board and BOOM: one of the transistor pins broken  . So I taken both, and then changed the connector. no problems there. So I bought transistors again 2n2904A and 2n2218 (the manual just specify NPN and PNP transistors and those were installed). So mounted stuff again and... Generator stopped to oscillate. Also it heats a little bit more than normal. All supply signals look OK, but oscillant signals are down or just with a DC level frozen. Disconnected transistors again and most signals on the board look ok. Obviously: none output. So I need a little help to repair this. I have the Original manual and errata. So I can check signalsinmediately, But I don't know where more to check what is the problem.

Thanks for your help 

[Whales]

Try replacing the big NPN and PNP transistors with cheaper/commoner plastic ones (eg 2N2222, 2N2907) on a breadboard.  You might just have duds/fakes or ones with an unexpected pinout.

[coldfiremc]

this is  .

I dismounted the NPN transistor from chassis but still connected and the generator worked. Without any distortion or saturation, all signals OK. I have to put an isolant there to prevent the short circuit?

[Gyro]

Try replacing the big NPN and PNP transistors with cheaper/commoner plastic ones (eg 2N2222, 2N2907) on a breadboard.  You might just have duds/fakes or ones with an unexpected pinout.

The output transistors (TO5 can) in the HP3310A are mounted to a chassis extrusion using heatsink clips and flying leads. It puts out quite a high voltage swing into a 50R load.

[Gyro]

this is  .

I dismounted the NPN transistor from chassis but still connected and the generator worked. Without any distortion or saturation, all signals OK. I have to put an isolant there to prevent the short circuit?

It sounds as if you've lost a (mica?) insulating washer that electrically isolates the transistor heatsink clip from the chassis.

[coldfiremc]

Fixed it with washers for both screws. Worked great, and transistors don't get hot now  . Could this be a definitive solution? or to put mica between clip and chassis?

[Gyro]

It sounds as if you've got a working fix. I can't remember (without taking mine apart) exactly how the transistors are insulated. I'll try to take a look tomorrow.


P.S. It's still a great little function generator, definitely worth the care and attention.

[coldfiremc]

they hadn't the classic switching distortion at peak that generators like Leader's LFG or goldstar had, and has a little higher frequency range than most cheap generators. It makes home-lab in company with a tek465M, tek2213, and an 6203B HP power source. If you can get information about mica, please post. I'm reading manual and I can't find anything yet

[Whales]

Try replacing the big NPN and PNP transistors with cheaper/commoner plastic ones (eg 2N2222, 2N2907) on a breadboard.  You might just have duds/fakes or ones with an unexpected pinout.

The output transistors (TO5 can) in the HP3310A are mounted to a chassis extrusion using heatsink clips and flying leads. It puts out quite a high voltage swing into a 50R load.

Yep, I was suggesting this as a test.  Having a breadboard inside your signal gen is not a long term solution either.

[Gyro]

It sounds as if you've got a working fix. I can't remember (without taking mine apart) exactly how the transistors are insulated. I'll try to take a look tomorrow.


P.S. It's still a great little function generator, definitely worth the care and attention.

Ok, I've taken the covers off to take a look. The electrical insulation (yellow) is built into the heatsink clips, no Mica washers involved. The two mounting screws are fitted with 'star washers' however. If these are missed out then I suspect that the tips of the screws can go in far enough to touch the tops of the transistor cans, causing your short circuit. You mentioned washers, so maybe this is the fix you already carried out. I noticed on my unit that one transistor isn't seated as far into the clip as the other one, maybe coincidence but I'm not going to touch it.

One other thing while you're in there. Check the condition of the three Rifa MP capacitors soldered directly between the pins of the mains IEC connector, if they show any signs of cracking then remove them! Maybe snip them out anyway. The X cap (L-N) on mine failed and caught fire a while ago - you can still see some remaining heat damage to the pin connector sleeves on the PSU board after clean-up. I haven't bothered to replace them with MKP types as the space and clearances are bit tight there anyway.

Photos attached. I hope this helps.


P.S. One final standard warning... The ramp trigger circuit uses Tunnel diodes for fast switching - CR21 and CR22 on board A1. These are sensitive and virtually impossible to obtain replacements. Be very careful if probing around in that area!

[coldfiremc]

I've taken 2 star washers from scrap. Mine is "heavily" repaired so it hasn't any socket for transistors. I've taken those 28awg cables with female terminal and put transistor pin there. on the other side, i retired plastic from terminal and opened the metal a little, so now the board pin matches with terminal. I don't know how this can affect middle or high frequency performance.



this unit still has those smelly sprague caps is also recommendable to change them?

[Gyro]

Wow, yes, your unit has been 'somewhat' repaired. A little bit surprising as the output stage is pretty robust and it has the engineering change to add output clamping (the two large stud diodes with red and blue wires on the aluminium bar).

I don't think minor modification of the connector is going to to make any difference to frequency performance.

I'm more concerned by the length and thickness of the wires going to the output transistors. I think they are long and inductive enough to affect the waveform at the 5MHz end of the frequency range, and probably cause crosstalk with other components on the PCB. The thickness is an issue for stress on the transistor legs (you mentioned that you broke one at the start).

If it was mine, then I think I would try to use thinner short leads as pictured in my original state unit (the missing sockets aren't an issue as there are only 3 wires to solder on each transistor). You could probably even get away without heatshrink with short leads.

I would also try to return that ground tag to a solid chassis grounding point with star washer, rather than sandwiched between two anodized aluminium surfaces with heatsink compound.

Good to see that the Rifa X and Y caps are already missing, actually it looks as if they were never fitted on your unit.


EDIT: Photo of the rear of the BNC connectors added, does yours have the 8R2 resistor on the main output connector? Maybe it is missing from your unit, causing the connection problem?

[Gyro]

One more thought. You mentioned that you have the manual, but not whether it is an original paper one. You might be best to refer to the latest (very final  ) pdf version if you aren't already...

https://www.keysight.com/en/pd-3310A%3Aepsg%3Apro-pn-3310A/function-generator?pm=PL&nid=-536900197.536896854&cc=GB&lc=eng

You've got to love HP /Agilent / Keysight for maintaining this archive for so long!

[coldfiremc]

it has resistor. don't knof if it's the original. attenuated output has a burnt resistor. Is there a safe way to replace it?

[Gyro]

It does sound to have been in the wars a bit.

It depends what you mean by safe. Access wise removing the front panel would give most space but also most pain in terms of disconnecting many things.  If you cover the PCB with something like thin card, to catch solder splashes, then you might be able to do the resistor change in situ with fine nose pliers and narrow soldering iron. also unplug the wires near them to avoid melting the insulation.

There are only two resistors in the attenuator - the series 1580R (call it 1k6) from the feedthrough insulator to the BNC center pin, and 51R1 (call it 51R) from BNC center to the ground tag. I assume that it's the 51R that is burnt through abuse.

Before changing anything, check the resistance across the attenuated output BNC connector - if it is near 50R then the resistor is probably just 'cooked' rather than burned-out. In that case, just leave it as is.

[coldfiremc]

looks like this 

[Gyro]

Ouch! That's been in the wars A LOT! 

Only something like mains applied to the attenuated output would cause that much scorching.

It's probably fair to assume that the 50R resistor is blown. Probably the easiest one to replace anyway as it's simply center pin to chassis tag. It's easiest to snip out the body (corpse?) and solder the replacement to the wire ends. Check that the 1580R resistor (R5) is still somewhere near the correct value too. The blown resistor (R6) is listed as metal film 51.1R 1% 1/8W but [EDIT: 50R 51R] is fine (1/8W or 1/4W). That doesn't matter as the level control is a variable pot, not stepped attenuator.

You probably want to clean the soot off the board with cotton buds and IPA (isopropyl alcohol).

[Tomorokoshi]

I have both a 3310A and a 3310B. Between them they had both problems shown here, a shorted Sprague capacitor and a burned out termination resistor. Those were relatively easy fixes.

A really useful feature of these that sets them apart from many other signal generators is the nice VCO range. With the right scaling it works nicely for simulating machine systems before the actual hardware becomes available, as long as one can work with the somewhat quirky frequency ranges.

[coldfiremc]

Nice info 

It's advisable to use thin coaxial cable to bias transistors? or just thin and short cable?