It helps if you mention what you are referring to exactly 😕
If you're referring to the original ExtremA amplfier it is set up for 2.5A of bias current, which yields 100W/8R in class-A due to the bridged configuration.
Reports about upgrade the ExtremA.
I'm working on upgrade the amplifier according to BV instructions. I'm testing separately front end and power output stage, because the amplifier has been unstable if the parts were connect together. The parts are working correctly, if they have been separately. Probably there are fights between the local NFB in power stage and the global NFB in amplifier.
BV and PMA, thank you very much for your support.
Cheers,
Bohdan
I'm working on upgrade the amplifier according to BV instructions. I'm testing separately front end and power output stage, because the amplifier has been unstable if the parts were connect together. The parts are working correctly, if they have been separately. Probably there are fights between the local NFB in power stage and the global NFB in amplifier.
BV and PMA, thank you very much for your support.
Cheers,
Bohdan
@Bohdan,
I would suggest starting with the less intrusive modifications I've suggested, that means four caps need to be changed out for a different value.
I would suggest starting with the less intrusive modifications I've suggested, that means four caps need to be changed out for a different value.
I agree. It is a bit irresponsible to suggest so many changes as BV did, based only on simulation.
Yes,maybe.. But You changed whole frontend, brigde configuration to SE and slow MJL instead fast Sanken´s. Is not it bigger change? I am not responsible for slow progres and for mistakes during tuning, this is a "remote controled" proces.
My target was to modify this amp to work reliable, with proper SR and stable DC behaviour, with unproblematic clipping behavior ..It was to many mistakes in original circuits, it is not possible to cure this all only with changing value for several capacitors.
My target was to modify this amp to work reliable, with proper SR and stable DC behaviour, with unproblematic clipping behavior ..It was to many mistakes in original circuits, it is not possible to cure this all only with changing value for several capacitors.
Last edited:
Proper SR is fixed with the changes I propose, I've got confirmation from two people already that implemented these changes, which matches the results in the simulator.
I'm not aware of any DC stability issues. It obviously has a DC offset trimmer fitted for a reason, set that with inputs shorted to GND after about 30 min warmup. However if you still see a shift in DC at the output then your source is likely the culprit here, make sure it is free of DC errors.
Clipping behavior is also fixed with the changes I propose as the loop is far more stable, hence clipping is also much better behaved.
I do however question what you are driving with this amplifier if you run it into clipping often, that means you're driving >100W into 8 Ohms. Take note that only 50W is available in 4R and even less at lower impedance due to the principle of operation of a class-A amplifier.
DC gain = 47k/2k2, equal to AC gain (without input coupling capacitors). Why??
Result-bigger thermal drift, sensibility to matching some transistors (input dif. stage, folded cascode transistors), ,sensibility to precise adjustment.
Clipping- inspect currents in CFP buffer after VAS stage, even without output load and input voltage overdrive (what is possible in real use).
It is basically a power differential opamp, so of course AC gain = DC gain, it would be silly if that wasn't the case. It goes without saying you need to use 1% resistors in the feedback network, and we specified that for best DC stability the input pair needs to be matched and the DC offset needs to be trimmed after the amplifier has warmed up.
If you are worried about a DC shift, especially with sources that are not free of DC errors then it is trivial to tie a DC servo around the outputs, a simple OPA277 opamp, a few caps and a few resistors is all you need.
If that's not something you fancy, and your source has DC errors, you're welcome to use coupling caps of course.
Having said that it is clear that such issues need to be corrected at the root, i.e. the source that puts out DC.
The piece we have spoken about has had horrible output DC fluctuations without any signal source, just with input terminals shorted. There is a kind of instability inside, difficult to find. My strong suspect goes to output stage local instability initiated by transients from supply diode switch off and transient recovery voltage spikes, through rails. Based on further measurements.
Last edited:
I can't really comment on something I've not seen myself but such behavior could certainly be attributed to the loop stability, which has improved with the suggested change of capacitor values. It is not outside the realm of possibilities that on some builds the loop stability is marginal due to larger than expected variations in parts or wiring/hookup.
I'd be happy to try and address any such issues with anyone experiencing DC fluctuations on the output. Just post your query in this topic and be sure to include detailed photos and a description of your build.
One known issue with DC offset I am aware of was due to the fact that a few people used counterfeit 2SA1216/2SC2922 parts. These parts also failed after a short period of use as they couldn't cope with the dissipation.
An externally hosted image should be here but it was not working when we last tested it.
I personally only purchase these transistors from a official Sanken distributor, which is also listed as such by Sanken themselves. I've had many bad experiences with the likes of eBay, Reichelt, ASWO, Nedis, etc. etc.
The fakes are such good copies physically, down to the texture of the case, the printing/font and the plating on the metal parts that you really need to sacrifice a pair and crack them open to inspect the actual die.
In respect to your photo, I can't say for sure, it depends on where these were sourced from.
P.s. there's an excessive amount of thermal grease used, and by the looks of it mica pads? I would advice against that strongly, as the thermal impedance is far too high. Thin, silicone, thermal pads should be used, Bergquist for example has good pads with great thermal characteristics.
Please keep in mind that dissipation per transistor is approx. 50W, that's a lot of heat to dissipate from an area the size of a poststamp, hence making sure the thermal resistance is low enough really is a requirement. Otherwise thermal imbalance will also result in higher DC offset.
Bergquist Sil-Pad 1200 is what I've used often and which works very well, this can be bought from various suppliers in large sheets, so can be cut to size easily with a knife/scissors.
Sil-Pad Products ~ Thermal Materials, Thermal Solutions ~ The Bergquist Company
Most thermal pads of the type you mention have HIGHER Thermal Resistance than thin mica with thermal paste on both sides.
- Status
- Not open for further replies.