John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
The 2 stage phono approach can offer several subtle advantages:
It can operate open loop, rather than like an op amp.
The phono EQ can be SPLIT for greater accuracy.
The feedback resistor to the input can be eliminated, allowing VERY low input noise even without a transformer.
The dynamic range of the input signal can be greatly extended.
Probably more, but I can't think of any, right now.
We selected it originally in 1979, by listening tests, doing an A-B experiment while I was working at HK.
 
Last edited:
Disabled Account
Joined 2004
The feedback resistor to the input can be eliminated, allowing VERY low input noise even without a transformer.

I don't think you could ever beat the DNR + noise of a GOOD transformer, and they are not obsolete (better yet in a dire need you can find someone to fab a new custom one). Also typically very low order distortions that vanish at low levels.
 
For tubes, they are mandatory, because the self noise of a tube will seldom go below 200 ohms.

For D3a in pentode, ENR is about 150 ohms. In triode, it's about 60 ohms. SS will do somewhat better, but with an input transformer, the input-referred ENRs end up looking like single digits, not to mention stellar CMR. And of course the tubes will trounce SS for open loop linearity and overload margins.
 
Well, to each, his own. SY and SE just love transformers, cost effective ones too! Perfect in every way, I guess. I'm stuck with using more expensive (and perhaps exotic) step-up transformers from Lundahl. Got two sets here, not sure which will sound better (if that is at all possible that there be an audio difference between audio step-up transformers) The two I am comparing are the LL1931 and the LL1933. Not cheap.
 
Well, I am usually stuck, these days, with jfet discrete devices or low noise IC's that are necessarily bipolar transistor input. For the last decades, about 1nV/rt Hz is available with a differential pair of input devices. Darn good. The same can be made with Toshiba jfets, but with the additional advantage of less input current noise, that ALWAYS limits the use of a bipolar input and the use of a transformer, if it is necessary.
45 years ago, we used input transformers, especially for audio tape recorders, and tube based mixers. The transformers NORMALLY supplied were not the best, but the budget models, and they were audible to some extent. In 1968, I worked hard to eliminate the NEED for an input transformer in solid state tape reproduce stages, as the unit we were using that cost at an OEM (contracted bulk purchase price), the amount, a typical engineer might earn in an hour, had lots of distortion and it gave an added noise contribution at high frequencies, due to eddy current noise generation due to the relatively thick laminations used, by BEYER, the transformer manufacturer. When I confronted the manufacturer about this, they said they could do BETTER, but nobody asked for it before. This is when I started using low noise jfet inputs, and paralleled, relatively low Rbb' bipolar transistors such as the 2N4405, to facilitate removal of the transformer in future designs.
Later designs from just about everybody, tended to remove the input transformer from the equation.
Still, most phono cartridges of the day, except for Ortofon and maybe EMT, tended to NOT need input transformers, and adding them would NOT have improved the S/N in any significant way. However, the Ortofon input transformers were WORSE than the BEYER/Ampex transformers in every way, due to cost considerations, I would presume. That is WHY I started to work on an MC solid state replacement, based on my work for Ampex, beginning in 1968. It took 5 years before I got somebody to take my design seriously. That was Mark Levinson, and we built the JC-1, transformer-less pre-preamp in order to replace the Ortofon and other input transformers, and this became the start of our mutual popularity in the audio marketplace. And so it goes.
 
Great transformers are expensive. More expensive than most solid state inputs. Of course, good transformers are somewhat cheaper. For tubes, they are mandatory, because the self noise of a tube will seldom go below 200 ohms.

On (R.I.P.) Allen Wright's website, there are MC stages without transformer, input is JFET cascoded by tube and second stage is tube only.
 
This is usually a quieter topology than a single tube for 2 reasons. First, even with equal Gm/Id between the jfet and the tube, the jfet statistically has a lower noise, due to the fact that it operates at near room temperature internally. Tubes run hot, inside.
The second reason is 1/f noise, and tubes have it in abundance, if you are not careful. Good jfets can have very little 1/f noise in the audio or at least above 100Hz band.
The number cited in an early Ampex AES paper from the early '60's, is 2.5/Gm for tubes, .5/Gm for bipolar transistors, and .7/Gm for jfets to calculate the equivalent input noise for the device. You may do the calculations yourself.
 
Disabled Account
Joined 2004
SY, I seriously doubt that your transformers cost more than the transformers I am using.
Are you using the 9570 with OCC wire option? If not, what model and construction?

If I (could/did) order Toshiba JFET's from Erno's website they would cost a lot more than those Sowters. Sowter is refreshingly honest making the statement that there is no scientific evidence for the OCC wire.
 
Even single diff is not too bad:
 

Attachments

  • singlediff.PNG
    singlediff.PNG
    33.8 KB · Views: 269
Status
Not open for further replies.