Amp with high pitched static noise

[Radio Wrangler]

If you can, support the leg close to the transistor's package. Bending a leg has a lot of leverage on the plastic to metal sealing where the leg enters the plastic. this can craze the plastic allowing moisture in by capillary action. All too small to see, but over a period can result in destruction of the device. If trimming leads, be very gentle with cutters. scissor action ones aren't bad, but the nipper type squeeze the metal until it fractures. When this happens, there is a lot of tension in the nipped metal as the nippers try to stretch it. When it finally splits in two, it creates a shock wave in each direction - along to the end of the cut of part, and along, into the transistor.

Doing serious electronics and tracing all failures back to cause shows you that making things be reliable involves lots of little acts of care. Nipping things off with your finger supporting the nipped-off end dissipates some of the energy and spoils the shock waves.

The last firm I worked for did aircraft radios for talking, navigation and instrument-landing as well as the transponders to link aircraft into air traffic control and automated anti-collision systems. For some reason everyone wanted everything to be very reliable!

While working at HP we got interested in reasons for in-warranty failures and instituted a programme to reduce our failure rate by a factor of ten, and set a target of getting it done and proven within a ten year period. In the end we did a lot better than that and what we did improved reliability right the way beyond the warranty period. We weren't fixated on warranty because of cost saving results, it was just that we got the best info on failures when they happened during the warranty period. People are keener to get things fixed when they know they're not paying for it.

Careful handling, anti-static precautions etc aren't just people being pernickety, it resulted in products with about 1/20th the number of breakdowns they used to have, and we all want some of that.

david

[steve ss]

Ok!

So a quick look.

You have no lack of experienced techs or good advice here but you do lack time!

You need to buy some in the form of a variac.

As you can then vary the rail voltages at will keeping them below the destruction

point of the Fets etc. and giving you time to measure and locate the probable source
of the fault!

You are fortunate to have one working channel to do the comparisons against.

As already mentioned you can make up a temp heat sink using folded aluminum

sheet (say 1"X4" strip folded in half) attached to a bulldog clip that will also buy you
some time.

Swapping components from one side of the board to the other can result in mirror
reversal of the component leads "not good"(all ready mentioned to ).

As you have one working channel it will be possible to ascertain at what minimum voltage the amp will function and if you are lucky it will be low enough to get some
audio through the faulty channel once you get the amp "in near working condition" and may be a guide as to the suitability of any substituted parts without any component failure.

Next add a temporary blocking/coupling capacitor (say 100mfd 100volt )to the amp output (faulty channel) to prevent DC passing to the speakers.

With lower rail voltages the risk of oscillation is usually decrease also.

I see you mentioned a light bulb tester(used as a current limiter) , often useful,but not a suitable substitute for a variac in this case.

I have seen them (variacs) on ebay new and used and some were quite affordable.

The alternative is to borrow one if possible!

or buy one use it for this project and then sell it on.


So with the tech help from the team here this should get you over the line!


I hope some of this helps!


enjoy Steve ss.

Ps. When you "win" you may have buy them a round Cigars!

[spacebiscuit]

An update:

I had been tentatively powering up the amp until this point with one new 2SJ352 and 2 x J2221 (one original one new) which was giving me hot FETs all round.

Yesterday I wanted to establish if the other original J2221 I had was good - better to use 2 originals if they work. So I swapped it the new J2221 so I had the 2 x original J2221s and one of the new 2SJ352. Initially I had sound, I then left the original J2221 just fitted and swapped the other (known good) out for the new J2221. When powered up I had only very feint sound output which was distorted so I began to fear the worst and thought it might be time to put an end to all of this with some cricket practice using a hammer and the amp.

Before I lost my temper I had the sense to remove the 2SJ352 and was relieved to be getting sensible readings from my component tester. I knew I also had one good original J2221 which had been removed. This good known pair was put to one side.

With the new 2SJ352 removed I thought I have nothing to lose, so put one of the first replacements 2SJ352 on the board - the one which tripped the fuses. When I powered up my bulb stayed lit and speakers hummer. That FET now has a big cross on it in marker pen. Cheap low quality junk.

So next I removed all J221s and put the one known good working one on the board along with the known good 2SJ352 plus the other new replacement 2SJ352 (until this point still in packet - I know - face to palm!). So we're now testing 1 x original J2221 and 2 x new 2SJ352.... sound was great (same as working channel) for the 15 or so seconds that I tested - and the 2SJ352s were no where near as hot as they were previously (still not bolted down to heatsink) - presumably they work in tandem to share the load.

Next up I will try and add the missing J2221 - I have 2 from a previous bath of unknown quality and one from the latest batch (seemingly better quality) - so I'm still alive just (I think).

Having read above that contributors working at HP working on reducing fault makes me only feel more like an imposter here and I am sure my approach to "testing" isn't impressing anyone.

I'm sure I will fit 4 working FETs and it will all go up in smoke after 30 seconds - but until then there is hope which is keeping this project alive.

Thanks for bearing with me thusfar.

[Analogue man]

I think the heatsinks must be fitted before you try to power up the amp again, this will at least give the FET's a fighting chance against thermal runaway, but as I said in post 160 don't forget the insulators otherwise good bye FET's and more fuses.

[Diabolical Artificer]

Sounds like you're making progress. If it works, great, if it don't at least it's kept you out of trouble for a few weeks, better than vandalising telephone boxes. You hopefully know a bit more now than you did too, knowledge is valuable.

Andy.

[spacebiscuit]

So I have established that I have 4 working FETs at least.

My next step is to bend those pins and screw them down to properly test. The one positive about the junk fakes I bought is I now have a few which i can use to practice pin bending with!

[Diabolical Artificer]

So I have established that I have 4 working FETs at least.

Progress, good.

The one positive about the junk fakes I bought is I now have a few which I can use to practice pin bending with!

Good thinking, I was going to suggest that but.

When you have it back together and before you play music get your meter, set it to 20V DC then take a reading at the speaker jacks. Adjust the meter to a lower setting - 2V ? same again. Do this for both channels.

Andy.

[spacebiscuit]

realised something - I as being lazy and plugged my soldering iron into my bulb tester - it the temperature is set above 300 the light comes on and dims as I turn it down. Even at the temperature at which the light has dimmed the iron struggles to melt the solder, plugging it direct by-passing the mains same temperature and no issues.

Presumably the light build tester is taking some of the load - would this suggest that my iron is drawing more current than it should / out of spec?

[Diabolical Artificer]

Presumably the light build tester is taking some of the load - would this suggest that my iron is drawing more current than it should / out of spec?

Yes to the first part, it's in series with your iron/amp,that's why we use them.

Second part, no idea. You really need to get your head round ohms law. Current voltage resistance, the basics. Without understanding that your flopping about like a beached fish & most of what we say, try to teach you will be meaningless.

Basic stuff, 4 stroke engine, boiling point of water, hydraulics: without a basic grasp of this basic physics your missing a lot, walking blind. Apologies for the preaching, but you need to do some work too, get the old noggin working.

I know it's hard, I struggled too, but with the help of fellow members of this forum & some reading, I started to figure some of it out.

Andy.

[spacebiscuit]

So I soldered the FETs into place properly, bolted them down with thermal pads and a thermal paste between.

On a couple of occasions when I powered it up and I had a bit of a hum when my source was turned right down, this then disappears as I turn the source up until it disappear completely. Other times that hum isn't there - but I need to ascertain if it only happen on one of the speaker outputs (I tried both). Likely sound stupid, but almost as if it is warming up!

I gave it a good test for about a minute or so, I pushed it a little but not excessively, sound is great. Before bolting the FETs down I was touching the back to feel the heat, now obviously I am unable to do that as the back is on the heatsink, the bolts on the underside that anchor the FET down are warm to touch but not on fire so maybe the heatsink it doing its job.

Not really sure what to do next, I'll read back over this thread as there were some readings it was suggested I take. I know I'm not quite out of the woods just yet, but a small achievement it seems.

[spacebiscuit]

Took a reading on the speaker jacks, good channel 0v, repaired channel 0.15v but seems to be fluctuating. The sound was really good when i tested this morning and left it running for around 20 minutes - no feedback or unwanted noises.

I thought about maybe adding a fan to the case to keep things cool - I'm super paranoid that it is warm but it could have been like this previously. As mentioned back in the day I used to place a fan in front of it when I was driving it hard.

[Diabolical Artificer]

Really chuffed the amp is fixed, it's taken a long time to get here. I hope you got a buzz to have got this far. The OP DC readings seem to show you have no bad DC offset. I think there maybe still a bit of an issue, but hopefully it's a minor one.

Compare the two channels: is the temperature of the two heatsinks about the same? One method we employ to fix amps is to take DC readings of the two PCB's/channels and compare. I'm not sure that's a good idea here as the potential for a slipped probe could cause damage.

Your amp circuit is a bare bones & simple, as you say it's always ran hot. Far better more powerful PA amps can be had for not much cash. If you intend to use your amp in PA service I'd advise getting something better. If you want to use it for light home use, you may be ok.

Hopefully this thread can now be closed, and marked up as a successful repair, Andy.

[spacebiscuit]

Yes absolutely I'm over the moon - I want to say a huge thank you but I'm reluctant to do so for the same reason I haven't yet put the amp back together, because the moment I do so I feel like I'll be tempting fate and it will go bang!

So yes it seems like a successful repair, had never heard of MOSFET before this experience and if I am honest I probably thought those black tabs were simply anchor points to hold the board down. I've made myself a bulb tester, became the proud owner of a component tester (already broken), and my soldering skills have improved.

The amp is mostly retired now, I use it in my summer house so is total overkill for what I need it for but this experience had made me kind of fall in love with it, more than I ever did before - it was just a tool previously. I have a real appreciation for the sound it produce too - rich deep and full of body.

Well it only took almost 10 pages to get here. Did I mention that one of the led channel level displays doesn't seem to work? I'm just about to remove it because the spade terminal needs resoldering - even less to this board so it can't be a difficult fix can it?

[Radio Wrangler]

Well, it's been a great learning exercise for you. It's a pretty stripped-back amplifier and not a good example of what I'd think of as a good reliable long-term prospect, but what you've learned and the confidence you've acquired is likely to long outlive it.

Those power MOSFETs burst on the scene 45 years ago and a lot of people went ga-ga over how much easier amplifiers seemed to have become and how little hardware you needed. It took a while before it was appreciated that less gung-ho designs had better staying power.

Congratulations!

David

[spacebiscuit]

Originally Posted by Diabolical Artificer

Compare the two channels: is the temperature of the two heatsinks about the same? One method we employ to fix amps is to take DC readings of the two PCB's/channels and compare. I'm not sure that's a good idea here as the potential for a slipped probe could cause damage.

I only have one speaker on my bench to test with - they are very bulky to move so I'm only ever testing one channel at a time. The new channel does seem to run a little warmer. I have probes with retractable hooks so I can safely take any reading you might suggest without danger slipping probes.

Just had a look at the LED display. I had to repair the solder joint where a spade terminal came loose. Both channels work but I have to play the system very loud just to have the first of 12 leds to illuminate. There are what look like POTs on the board, I tried adjusting these but it only disabled the led completely. Presumably a resistor is at play here - R4 and R12? Could these be changed to increase sensitivity? Images attached.

[spacebiscuit]

Referring to this thread where with the help of the forum I managed a repair to a channel with blown FETs:

As mentioned in the conclusion, when I first power up there is some oscillation, at least that is what I think it is - a deep warbling sound. I power the amp down and then back up and play a source at it stabilizes.

I read that an option might be to add 10pf-100pf caps soldered between Gate & Drain on the FETs?

Any thoughts on this idea? Could it help?

Thanks.

[Diabolical Artificer]

Does the other channel do this? A "deep warbling sound" sounds like low frequency oscillation which means adding 10p caps from D - G might not help, this being a cure for HF oscillation. HF osc should be taken care of by the zobel network, the 20r & 220n cap across the speakers. Same goes for C5/R7.

If the other channel is ok, then this osc is down to either a fault or the new fets. Without a scope t& a lot more time the first is tricky to diagnose. you can try adding the caps from D-G but I'd trying changing all the 3k3 gate resistors after having confirmed they all ok & that isn't the fault. Check all the source resistors too are ok, one or two may be OC or have gone high.

Others may have better ideas, Andy.

[spacebiscuit]

The issue only exists on the channel with the new FETS, it didn't previous exist on either channel prior to fitting.

Regarding fitting the cap between G and D on the FETs - any harm in trying this. Somewhere else I read this was not a good idea - as long as it can't do any damage I'd be willing to give it a try unless told otherwise?

[Radio Wrangler]

There's only one way to find out for sure.

BUT the devices are quite fast devices internally, and get slowed by stray coupling capacitances, connection inductances (the bond wires from chip to lead-frame) etc. They can indeed go unstable, using their stray C and L components as coupling factors making up a stray circuit not too far from a Colpitts oscillator circuit. There's enough gain for oscillation or marginal oscillation. What you've heard could be a too-high-frequency-to-hear oscillation starting up and dropping out at an audible frequency.

On the whole these things are never good and can result in damage. As a minimum they use-up dynamic range and increase distortion.

You need an oscilloscope good to a few tens of MHz to check there's nothing wrong.

Would I try it?

Not in a month of sundays without some very careful checking and understanding first. Do a random test, get a random result. Only if you know what you did in full detail do you get to be able to interpret the outcome with any certainty.

David

[spacebiscuit]

Originally Posted by Radio Wrangler

You need an oscilloscope good to a few tens of MHz to check there's nothing wrong.

Would this help me?

https://cdn.rohde-schwarz.com/hameg-...07_english.pdf