In many Class A/B amps the THD+N curve vs output power starts higher because of small signal. Even if grass and harmonic spikes are low, its about a ratio. Thus the curve goes down as the signal output increases. Sometimes flattening out sometimes creating a shallow valley. Until taking off steeply when near the clip level.
"Just now I have done a loop test"
Sound card line out to sound card line in.
Results are the same.
Distortion decrease at higher level out.
That is the case for the sound card too.
It is typical behavior of push pull with a bit of bias and much NFB. Operational amps are small amps after all. There is a sweet spot of THD performance in any sound card. Usually between 0.5V to 1V RMS input level for affordable home studio grade models. If feeding the card attenuated signal via a few kΩ pot wired across the power amp's dummy load you can keep the ADC input level steady if careful. Both for the card's safety and to can plot manual THD vs Power curve.
This requires a volume pot. in amplifier input then.
And the adjustable attenuator across the dummy load.
Then you can keep the sound card volume controll in the same level.
Last edited:
Sound card usually has a big knob volume output pot. Useful to quickly control the signal level going to the device under test. In case it misses such a knob (as in some older models) there's still the software's output volume control. Which is not that practical when performing bench tests but it does the job.
Yes but how to do this as soon as sound card used as out and input device simultaneously?
Say you want to keep the return to a sound card's input at 1V RMS when testing a 28dB gain power amp in its loop. That's 25X raw voltage gain. You start with 1V/25=0.04V i.e. 40mV RMS card output. Dummy load's attenuation pot fully open for now. You are at 0dBV (1VRMS) return. Adjust card's output accordingly for different sensitivity amps. Also have a DMM or scope hooked across the dummy.
Slowly increase the card's output. Don't let the FFT climb more than double (6dBV). Back down the dummy's pot for keeping the FFT at 0dBV (1VRMS). And so on and so forth.
What RMS voltage you see on DMM or scope derives the actual power (Vsquared/Rdummy). The portion you send to the card's input via the pot is to keep the card's own THD contribution small and steady at 1VRMS.
Slowly increase the card's output. Don't let the FFT climb more than double (6dBV). Back down the dummy's pot for keeping the FFT at 0dBV (1VRMS). And so on and so forth.
What RMS voltage you see on DMM or scope derives the actual power (Vsquared/Rdummy). The portion you send to the card's input via the pot is to keep the card's own THD contribution small and steady at 1VRMS.
I use Jan Didden's Autoranger attenuator and I had tested various amps without seeing a particular pattern due to the card and attenuator. Others start on lower THD+N and pick up with wattage others start bit higher and level off until clipped. Mostly different due to amount of bias, topology, tubes or transistors, etc. In general the amp is going to smother the card's contribution if you will keep return signal to card enough and moderate.
The basic topology yes, but for the design as it stands I would say 'not recommended'. The T092 transistors for the VAS and current source would be getting to the point of dissipating to much power for free air cooling. 45 to 55 volt gains you nothing in subjective increase in output.
That was not in the original design. Have a read here:
https://www.diyaudio.com/community/...lifier-designed-for-music.119151/post-5516200
Changes like that are 'untested' and while almost certainly OK from a 'will it won't it work perspective' the 'high gain/low feedback' philosophy was a core aspect of the design from the outset. How changes to that might effect the final presentation of the final sonic result is an unknown.
Ok, thanks for the answers. I understand that there are two feedback loops then and they are complementary, so the overall gain is the mix of both feedback loops. Therefore it is possible that the overall gain is right from the start for me.
I think that the manufacturer Densen in his old amplifiers used a negative feedback loop that he took from the output of the VAS and not from the final output of the amplifier, it seems something similar...
The output power at 8 ohms with +-45v I think will be about 80wrms... I have designed some amplifiers in my life, always with long tail Ltp (diff amp) and with exicon lat mosfets (they are the ones I like the most). I once tried a preamp with almost this same unique input bjt layout and I liked it a lot, I'm excited to build this amp... Thanks!!!
I think that the manufacturer Densen in his old amplifiers used a negative feedback loop that he took from the output of the VAS and not from the final output of the amplifier, it seems something similar...
The output power at 8 ohms with +-45v I think will be about 80wrms... I have designed some amplifiers in my life, always with long tail Ltp (diff amp) and with exicon lat mosfets (they are the ones I like the most). I once tried a preamp with almost this same unique input bjt layout and I liked it a lot, I'm excited to build this amp... Thanks!!!
The 680k alters the midband gain by only a tiny fraction of 1db mainly having more of an effect at frequencies well above the audio band... but still a very small effect.
All these are things that can be experimented with and tried but as I keep saying, they are all untested (by me anyway) modifications. The original design has proved itself over many years now and I can only really say to stick to the original concept. You won't be disappointed in the results
But don't let that stop you experimenting... once you have built the design as intended you then have a baseline to compare changes against.
That's a realistic figure.
All these are things that can be experimented with and tried but as I keep saying, they are all untested (by me anyway) modifications. The original design has proved itself over many years now and I can only really say to stick to the original concept. You won't be disappointed in the results
But don't let that stop you experimenting... once you have built the design as intended you then have a baseline to compare changes against.
That's a realistic figure.
An increase in supply rail voltages to +/- 55V would entail an extra set of MOSFET devices for reliable operation e.g. as per JLH's Audio Design project from ETI magazine in 1984 July issue. That would entail extra cost for parts which could be a problem of availability of the MOSFETs and a redesign of the pcb.
I
The 2SK1058/2SJ162 that I use were similar electrically to the older 2SK135/2SJ50 which were in a T03 package and I remember Maplin running those at -/+55 volt while claiming 100wrms into 8 ohm and 150 wrms into 4 ohm, all from a single pair.
I think that's pushing it tbh but the FET's themselves are OK at that voltage, the limitation with my amp is the dissipation in the earlier stages. It would need different transistors, possibly with very small clip on heatsinks for total reliability if it was in a confined enclosure.
I would agree though that if using a higher rail voltage then multiple pairs or the double die versions would be advisable.
I think that's pushing it tbh but the FET's themselves are OK at that voltage, the limitation with my amp is the dissipation in the earlier stages. It would need different transistors, possibly with very small clip on heatsinks for total reliability if it was in a confined enclosure.
I would agree though that if using a higher rail voltage then multiple pairs or the double die versions would be advisable.
The closed loop gain with a 22k feedback resistor and a decoupling arm to earth of 1k is 22 times and often that is not enough for phono stage amplifiers especially if the RIAA curve is dependent on a passive network or a passive network in addition to an active stage in a two stage design.
22 times is probably about par for a BJT output stage due to limited Gain bandwidth product.
These days turntables are being offered with on board phono stages so with high closed loop gain using MOSFET output devices .i.e double BJT capability, line stage ahead of the power amplifier can be dispensed with and switch selection and volume controls can be housed in one single case.
22 times is retrograde thinking to my mind - akin to going back to the limitations of push rod internal combustion engines, low octane fuel, and three to four speed gearboxes.
If turntable manufacturers have gotten into onboard phono stages let them compete with each other just as CD manufacturers etc have to do.
22 times is probably about par for a BJT output stage due to limited Gain bandwidth product.
These days turntables are being offered with on board phono stages so with high closed loop gain using MOSFET output devices .i.e double BJT capability, line stage ahead of the power amplifier can be dispensed with and switch selection and volume controls can be housed in one single case.
22 times is retrograde thinking to my mind - akin to going back to the limitations of push rod internal combustion engines, low octane fuel, and three to four speed gearboxes.
If turntable manufacturers have gotten into onboard phono stages let them compete with each other just as CD manufacturers etc have to do.