At some point, I'll be given enough free time to finish the last two chapters of the Heretical Preamp saga, but one result I got was so striking, I thought I'd toss it out in the meantime. As much resistance as I've seen to using input transformers (and I was certainly in that camp for a long time!), the advantages are many, as I've discussed in the first installment.
Here is a wonderful example. This is the servo version of the Heretical (next chapter), set up for a 1kHz THD run. I used two stereo mini-plug to RCA adapters to connect to the sound card input and output; and a 1.5m interconnect pair, with one channel run to the Heretical input, the other channel run to a female-female coupling (oooh!) attached to the sound card in. The other channel of the sound card in was connected to the Heretical output. So our comparison is 1.5 meters of interconnect plus the Heretical versus just 1.5 meters of interconnect.
The test was done in a horrible location- because of the short mini-to-RCA adapters, the preamp and cable were sited behind the computer, on the floor, atop a pile of active power cables and an open computer box.
The spectra are shown here- the output level was set to 1VRMS. The white spectrum is displaced upward 20dB from the green spectrum just for clarity. In reality, they both had a similar baseline, about -125dB.
Note how much crummier the green spectrum is.
Now the punch line- the white spectrum is with the preamp in the loop! Even with the cover off, laid on power cords, and everything else I could throw at it, it STILL meets the noise spec, with 60Hz hum being the worst component at -95dB ref 1V (that's -101dB ref 2V). I might modestly point out that the ripple is nearly indiscernable...
So although I like to say that a preamp can't improve the sound, at best it can only pass it along unchanged, there are special circumstances that provide the exception. My subjective reaction that this preamp is quieter and clearer than the one it replaced may indeed be a valid one.
If you build this, do use the input transformer.
Here is a wonderful example. This is the servo version of the Heretical (next chapter), set up for a 1kHz THD run. I used two stereo mini-plug to RCA adapters to connect to the sound card input and output; and a 1.5m interconnect pair, with one channel run to the Heretical input, the other channel run to a female-female coupling (oooh!) attached to the sound card in. The other channel of the sound card in was connected to the Heretical output. So our comparison is 1.5 meters of interconnect plus the Heretical versus just 1.5 meters of interconnect.
The test was done in a horrible location- because of the short mini-to-RCA adapters, the preamp and cable were sited behind the computer, on the floor, atop a pile of active power cables and an open computer box.
The spectra are shown here- the output level was set to 1VRMS. The white spectrum is displaced upward 20dB from the green spectrum just for clarity. In reality, they both had a similar baseline, about -125dB.
Note how much crummier the green spectrum is.
Now the punch line- the white spectrum is with the preamp in the loop! Even with the cover off, laid on power cords, and everything else I could throw at it, it STILL meets the noise spec, with 60Hz hum being the worst component at -95dB ref 1V (that's -101dB ref 2V). I might modestly point out that the ripple is nearly indiscernable...
So although I like to say that a preamp can't improve the sound, at best it can only pass it along unchanged, there are special circumstances that provide the exception. My subjective reaction that this preamp is quieter and clearer than the one it replaced may indeed be a valid one.
If you build this, do use the input transformer.
I'm currently on the umpteenth iteration of my PP linestage, and I am totally lost as to what constitutes accurate sound and what doesn't. I came into this stupid project thinking a linestage had the easiest job to perform, relative to a phonostage or power amp....but no, it doesn't seem to be so simple. All I can say is, my current state of confusion is the best sounding confusion to date.
...sorry if I hijacked your thread...but I do use input transformers/phase splitters.😉
Pedroskova: Well, that's the thing about optimizing circuits. This started out as such a simple cathode follower and suddenly it's sprouted satellite perfboards everywhere... I'm finding that with a solid design and attention paid to reducing the criticality of parts and environment, there's less to twiddle by ear; nearly all my listening-test time was spent moving the servo time constant around. Everything else just dropped into place.
Don't worry about the hijack- this thread is more about philosophy anyway.
Don't worry about the hijack- this thread is more about philosophy anyway.
SY said:
In part based on what I saw in your first installment a pair of Hammond 850 series small signal transformers were tried on the input of a single stage EL84 play amp in the hopes of a little passive gain. Turns out they weren't suitable to the application but some basic tests with an Audio Precision portable resulted in distortion readings so low it took multiple tries before I was convinced the test was being correctly run. Weighted THD was well below .01% at levels relevant to consumer audio across most of the band. Pretty amazing.
> in a horrible location-... behind the computer, on the floor, atop a pile of active power cables and an open computer box. ...with the cover off
> If you build this, do use the input transformer.
OK: if you do every possible thing wrong, a $20-$200 transformer can mitigate your sins.
OTOH: a $2 box-cover, a PC-cover you already paid for, and $14 of decent-length PC/HiFi cables (so you could get out of the snake-pit) might do the same, and without the softening/sweetening effects of the iron core.
I'm not really putting you down, because...
> there are special circumstances that provide the exception.
.... in another field, the Pro Audio racket, interfaces are almost invariably Transformer Input (or a lame diff-amp chip approximation: not the same, but helps a lot). They meet "special circumstances" all the time, at high hourly rates, and time futzing with grounding tricks or cobbling repeat transformers is bad news.
> Weighted THD was well below .01% at levels relevant to consumer audio
Driven from a low-Z source, a decent transformer worked below rating will run 0.01 or 0.001 (or better) THD at 1KHz. THD rises as level rises and particularly as frequency falls. Try rated level at rated low-frequency limit: a good tranny will show rated distortion there (and bad iron may be running 5%THD). The distortion is nearly pure 3rd with few higher harmonics, not too offensive. Some recording artists like to crank their iron to "add color".
BTW: if your real-world source is not low-Z, test the transformer AT the intended source and load impedances. Unlike good amplifiers (like some bad amplifiers), transformer performance is a lot about impedances. With a non-zero Z source, bass distortion rises and bass -3dB frequency rises.
> If you build this, do use the input transformer.
OK: if you do every possible thing wrong, a $20-$200 transformer can mitigate your sins.
OTOH: a $2 box-cover, a PC-cover you already paid for, and $14 of decent-length PC/HiFi cables (so you could get out of the snake-pit) might do the same, and without the softening/sweetening effects of the iron core.
I'm not really putting you down, because...
> there are special circumstances that provide the exception.
.... in another field, the Pro Audio racket, interfaces are almost invariably Transformer Input (or a lame diff-amp chip approximation: not the same, but helps a lot). They meet "special circumstances" all the time, at high hourly rates, and time futzing with grounding tricks or cobbling repeat transformers is bad news.
> Weighted THD was well below .01% at levels relevant to consumer audio
Driven from a low-Z source, a decent transformer worked below rating will run 0.01 or 0.001 (or better) THD at 1KHz. THD rises as level rises and particularly as frequency falls. Try rated level at rated low-frequency limit: a good tranny will show rated distortion there (and bad iron may be running 5%THD). The distortion is nearly pure 3rd with few higher harmonics, not too offensive. Some recording artists like to crank their iron to "add color".
BTW: if your real-world source is not low-Z, test the transformer AT the intended source and load impedances. Unlike good amplifiers (like some bad amplifiers), transformer performance is a lot about impedances. With a non-zero Z source, bass distortion rises and bass -3dB frequency rises.
PRR said:
If the trannie's got some headroom, that pure 3rd can be amoeliorated with just a wee bit of DC current in the primary or secondary. If you're driving the tranny with something that has a variable DC offset, you can just "dial in" what sounds best.
se
Sy,
I don't know, but your green graph is more than familiar to me. This is the usual plague which infects most of my soundcard measurements!
Because with the card output - input connection what You are creating is just a biig ground loop! This is broken by the transformer. [Thanks for the idea - next time will insert it 🙂 ]
An other way of breaking the loop is to use a CD player, as a source. [Burn the test tones to a CD, I have done that with RightMark, also]
I think if you would repeat this same test in this way [with the player], You would get more similar results on each channel.
But it's really a nice idea for the soundcard measurements, thanks again! [Why did not think of it myself earlier.. while in the job I did it several times [not in audio]]
Ciao, George
I don't know, but your green graph is more than familiar to me. This is the usual plague which infects most of my soundcard measurements!
Because with the card output - input connection what You are creating is just a biig ground loop! This is broken by the transformer. [Thanks for the idea - next time will insert it 🙂 ]
An other way of breaking the loop is to use a CD player, as a source. [Burn the test tones to a CD, I have done that with RightMark, also]
I think if you would repeat this same test in this way [with the player], You would get more similar results on each channel.
But it's really a nice idea for the soundcard measurements, thanks again! [Why did not think of it myself earlier.. while in the job I did it several times [not in audio]]
Ciao, George
The idea is not that this is the way to test things (it was a lame lash-up), but that the transformer allows even worst-case stuff to come out OK. And in equipment testing, you want worst-case.
Every time I see someone who has a beautifully neat setup with well-spaced components and perfectly run cords, I mentally congratulate them, then turn back to my own system with stacked components and lots of wires where they shouldn't be (out of practical necessity). If the preamp still meets spec in the snake-pit, the depredations of being stacked between a CD player and a crossover and being fed by sources with questionable grounds will not bother it.
I was just surprised at HOW effective breaking up that ground would be.
Every time I see someone who has a beautifully neat setup with well-spaced components and perfectly run cords, I mentally congratulate them, then turn back to my own system with stacked components and lots of wires where they shouldn't be (out of practical necessity). If the preamp still meets spec in the snake-pit, the depredations of being stacked between a CD player and a crossover and being fed by sources with questionable grounds will not bother it.
I was just surprised at HOW effective breaking up that ground would be.
I buffer the output of my computer soundcard with Jensen transformers before I send the audio to my power amplifier.
The improvement in fidelity is astonishing!
improvement in s/n is also dramatic.
The improvement in fidelity is astonishing!
improvement in s/n is also dramatic.
- Status
- Not open for further replies.