I know a few of you are in to hi-fi and/or electronics and I am trying to investigate an issue that I could do with some advice on.

An amplifier of mine has started to produce what sounds like a very loud 50Hz or 100Hz noise on 1 channel with the signal still audible through the same channel. Advice has suggested that maybe a power rail has gone down.

In my simplistic understanding of electronics "a power rail rail going down" suggests that one of the +ve or -ve sections of the PSU has gone down??

A little background:-
The amp is rather unusual in that it runs in Class AB (contrary to popular belief) but is biased to around 80W in class A at 8 ohm impedance with around 160W available in class B. The amp uses seperate PSU sections for each channel and continues to double power as load halves down to 1 ohm so around 640W or 1280W in class B (it is a bit less but it has a very large PSU). It has been likened to a welder in comparison to the majority of amplifiers.

My concern is how I rig up a sillyscope to the output, the amp has been known to turn regular speakers into flaming infernos when it fails and resmbling the opening sequence from the series of Bonanza as the paper cones go up in flames for those that are old enough to know what I mean! Connecting a speaker as a load therefore is probably not a good idea (the speaker connected to it during failure is rather unusual and now in storage/too rare and costly to to use as a dummy load.

Am I correct on thinking that obtaining some high wattage and high resistance (say 16-50 ohm??) and connecting a sillyscope in parallel will be the best plan?

If anyone has any suggestions on components to check I am all ears - my thoughst are checking the rectifying block and PSU caps, teh amp is around 30 years old and runs very very hot as a result of the class A operation so I am thinking of electrolytic cap failure due to heat as my first port of call but just swapping out caps for new is not really a sensible option as it uses 4 very large 40000uf 75V (75mm dia x 145mm H) caps which are around £80-£100 each.

A couple of pics:-



Additionally how do I check the ESR of the caps to see if they have dried out?

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R

Wow! these look like the SuperCaps out of the TS030!

_________________
Don
GT4DC Chairman
1994 Toyota Celica GT-Four ST205WRC JDM 269bhp @ 0.9bar
1994 Toyota Celica GT-Four Special GT 590bhp @ 1.8bar
1989 Van Diemen RF88/89 Formula Ford 1600
2008 Nissan Patrol GU 3.0L ZD30DDTi 154bhp

Can you confirm a few things first.

Is the 80w of class a the quiescent current flowing through the push - pull output transistors and not 80w of dc bias to the speaker.

is it single supply or + and -

is there a capacitor in series with the speaker ?

Use a resistor as dummy load.
check power supplies first.
most likely a transistor failed if it went suddenly, electrolytics usually go with a bang and debris, smoke, and goo.

_________________
If at first you don't suck seed, try drier grain.

80 w of class A, around 2A quiescent current through transistors at idle. Not a dc bias voltage applied to the speaker.

It has a seperate transformer per channel which feeds the high current rectifier block and 1 cap is connected to the +ve and 1 cap to -ve out of the rectifier block.

There is no capacitor I am aware of between output and speaker, either in amp or crossover of speaker although I did wonder if adding one for test purposes would stop any dc content (if present) due to failure to protect the speaker coil.

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R

That's what I expected but didn't want to make any assumptions. 2A is a lot, even 30 years ago negligable crossover distortion was possible with only a few milliams of quiescant. I suspect this was a 'no holds barred' type thing especially when considered with separate transformers for each channel and frequency response down to DC. I suspect the fan required to cool the resulting furnace makes far more noise than the harmonics would have

In view of the value of the speakers in queston, it may be prudent to add a capacitor in series for future protection. If you roll off at about 5-10Hz I doubt you will hear, sorry feel, the difference. Your neighbours might though . The downside is it will be a large capacitor and will need to be an unpolarised electrolytic or 2 conventional ones in series in opposing directions, each one double the required value. To calculate the value: C(farads) = 1/(2 x PI x F (Hz) x R (ohms)) so for 8ohm speaker and 10Hz C = 0.002F or 2,000uF (use 2x 4700uF in series with negatives connected to each other and ideally through about a 10K bleed to the negative rail to bias the capacitors)

Back to the original topic,
Measure the voltage across each power supply capacitor with a meter, they should all be about the same.

Using a scope with earth connected to the busbar on the capacitors, look at each power rail (you may need the scope on AC to see the ripple) The peaks of the ripple should be 10mS apart (100Hz). If it's 20mS then a diode will have gone in the rectumfriar.
If one has a lot more ripple (but correct frequency) it will either be capacitor gone high resistance or a lot more current being drawn from that rail.

_________________
If at first you don't suck seed, try drier grain.

Ace, thanks cdb.
Whilst the amp is a " no holds barred" design it is the smallest in the range with 4 larger models that each double the power! Even more overkill?

I do have some cheaper bass units I can use for testing purposes as a load but will use the capacitor as a safety net as I don't really want spontaneous combustion of the coils and cones as it won't help find the fault!

"Rectumfriar" lol

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R

See how you get on with power supply testing, then we can take it from there. Just use a resistor, or even no load for testing.

May be worth having a look at the amplifier output with the scope to see if there's a DC offset and what the noise looks like (50Hz/ 100Hz & shape) if you have a means of taking a picture of the screen (or saving to USB if you have a posh scope)

_________________
If at first you don't suck seed, try drier grain.

hmm, that didn't go to plan.

A little bit of the magic smoke was let out on the amplifier on my first attempt and on my second attempt the "dummy" load let out lots of magic smoke then went up in flames. I had kind of wondered if this may happen so a clear exit route, quick release wire connection and fire extinguisher was on hand.

I couldn't find my sillyscope (I think I lent it to someone a few years back!)

I wired up 2 dvms to record voltage across a single PSU cap and one set as a frequency counter across the rectumfriar. As a back up I set a sound level meter to record the frequency of the dummy load. The dummy load was a pair of old car audio subwoofer wired in series and rated at 300W and 5 ohm DCR (surplus so cheap enough to fry and cheaper than building/buying a proper dummy load of the correct rating).

On the first attempt the quick release speaker connection was a bit loose and hence no load was present and a brief spurt of magic smoke came out of a resistor near to the trannysister.

On the second attempt it was immediately apparent that the amp had gone DC and both dummy load speakers were propelled to their ends stops with a very faint 100Hz making its way through briefly before I could switch it all off and the smoke/flames started. (I am aware what DC does to a speaker so I turned it off as quick as I could muster). Lucky I had the SLM recording Freq as the sillyscope and dialling it in would have taken me far too long!

I think it is time to get an expert on the case and suspect it is transistor failure as CDB initially suggested.
The house still stinks of fried electronics!

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R

Wish I was there to have watched it.

_________________
GT4DC Events Co-ordinator
1996 ST205 UK

I hope you used Entry of the Gladiators as the test track


100Hz sounds like a post rectumfriar capacitator is incapacitated

no need for a test track steve - it was passing its own Ac signal superimposed on a very large DC element.

Thinking about it the frequency measured could be spurious, a 10" cone on its end stop will struggle to produce anything at all. Whatever it was producing would only be half the wave - either due to the speaker being on its end stop and unable to move any further in one of the directions or due to the amplifier signal. As the speaker was in free air it could also be that I was only hearing the harmonic of the noise due to frequency doubling and the cancellation around the cone of the signal (be it 50Hz or 100Hz).

I think the capacitor theory is likely as they do tend to get cooked in the heat of the amplifier and dry out the electrolyte, (presumably then becoming a short circuit and allowing DC through?) I am not sure if a transistor failing would also have a similar effect WRT producing vast quantities of DC at the OP?

Either way I think it will need a variac, sillyscope and a decent dummy load to figure out what has failed. Replacement of the capacitors (the small electrolytics in circuit not in the PSU) is a wise move regardless as they will be part fried (even though I believe that they have been changed at some point in 30 years!). The transistors and replacement of is something I will need to research as they may be difficult to obtain and ideally I would prefer not to substitute a similar part due to originality of the unit and possible impact on the equipment value. It is an old unit that has developed classic status and values have doubled in the last 5 years since I bought it. (probably as they are all failing now!)

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R

See if you can read the text on the transistors and I can have a look at availability.

if you have smoke escaping from resistors then something has gone short circuit in the amp. Power supply caps wouldn't cause that.

_________________
If at first you don't suck seed, try drier grain.

I need to do a search - the transistors are just labelled "Krell A" or "Krell AA"




Believed to be the correct Schematic with some transistors marked. I suspect that I will be better off replacing all (more likely to be a match if from the batch?). The cost (if available) will be negligible in comparison to the rebuild time required.

http://www.reelaudio.co.uk/KrellKSA50Service.shtml link for a part documented rebuild.

The PCB has plenty of scorch marks from heat - suspected to be from resistors. So I also need to look at how to clean PCB and probably remake a load of solder joints.

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R

Happy to have a quick look tonight if you bring it & circuit along and point to the resistors that smoked I can probably tell you which transistors are likely to be gone. I will also bring a meter and can probably confirm by measurement.

_________________
If at first you don't suck seed, try drier grain.

Ohh its one of them push pull a biased ab setups

I'd start by disconnecting the output stages and having a poke about with sillyscope and squiggle generator
IIRC there's a trick of connecting the output rails via resistors to allow testing without the Chernobyl effect kicking in