HPS 4.0 phono stage

[syn08]

Originally Posted by Bonsai

Syn08, any thoughts on the THAT matched pairs, or the MAT series devices in place of the types you used in your head amp?

For MAT02 (NPN) and MAT03 (PNP) I have measured Rbb at about 25ohm (slightly lower for MAT03), being in between the really low noise Hitachi and Sanyo FBET pairs and the 40ohm typical for "low noise" diffused devices. The big problem is the PNP beta. It doesn't go over 150, and that makes matching with the NPN counterpart very difficult, in particular if you look at the $25 price tag . But of course, if you are happy with a few microamps through the MC cartridge, and an overall 0.6-0.7nV/rtHz noise performance, they are pretty good.

THAT devices have about the same noise/Rbb as the AD parts, but beta (in particular for the PNP part) is very low: under 100. They are also significantly more expensive than the Sanyo FBET parts, not to mention the Fairchild.

The only advantage that I can think of is that these monolithic duals could probably be paralleled without balance resistors (saving a little noise), but I haven't tried this.

[Joachim Gerhard]

Great work !
Me being more of a bipolar guy really enjoy that circuit.
Your knowlage of active devices is quite staggering.
By the way THAT can make your own transistor if you buy 500-1000 pc.
My friend Jürgen Ultee has designed a phonoamp for Van den Hul with THAT transistors made to his spec. It is an interesting design with inductive equalization.
He is a trained HF engineer and knows a lot about inductors and layout.

[syn08]

Originally Posted by Joachim Gerhard

Great work !
Me being more of a bipolar guy really enjoy that circuit.

Glad you like it.

Actually, it's a more difficult and demanding construction compared to the HPS 3.1, already on my web site. This design should be considered more like a backup plan, in case one is unable to find the Toshiba JFETs (now or in the future). It has better noise performances, but at those levels HPS 3.1 is already good enough, for the most demanding ears, anyway.

[Bonsai]

Syn08, you have proably addressed these questions elsewhere, so if they are repeats, please accept my apologies in advance.

On the HPS3, you use a balanced/symmetrical JFET input that drives folded cascode type VAS stage. What are your thoughts on a straight diff amp input (maybe with multiple devices in parallel to improve the noise performance). This would drive a folded cascode, or a mirror arrangement and then onto the op-amp based second stage. Is the noise performance of this balanced stage better than a classic long tail pair (again, we are talking JFET here). I have to admit, the classic long tail would have been my 1st cut approach, especially since the PSSR of a long tail pair is better than a folded cascode arrangement.

BTW, I have partially built up a phono amp design based on Dennis Colins AX design, I've changed some of the values (notably lower resistor values and higher cap values) and used LM4562's. I have yet to finish it off - hope to do so in the next few weeks.

[syn08]

Originally Posted by Bonsai

Syn08, you have proably addressed these questions elsewhere, so if they are repeats, please accept my apologies in advance.

On the HPS3, you use a balanced/symmetrical JFET input that drives folded cascode type VAS stage. What are your thoughts on a straight diff amp input (maybe with multiple devices in parallel to improve the noise performance). This would drive a folded cascode, or a mirror arrangement and then onto the op-amp based second stage. Is the noise performance of this balanced stage better than a classic long tail pair (again, we are talking JFET here). I have to admit, the classic long tail would have been my 1st cut approach, especially since the PSSR of a long tail pair is better than a folded cascode arrangement.

Unfortunately, differential stages come with a 3dB penalty in noise. So, you need four devices in differential configuration to get the same noise as one device in single ended configuration. To add insult to injury, building an ultra low noise current source to feed a LTP is a problem by itself. True, it's noise is common mode, but will appear at the output due to the inherent finite CMRR.

[Bonsai]

"Unfortunately, differential stages come with a 3dB penalty in noise."

but does the balanced input stage also not suffer the same fate? Seems we cannot have our cake and eat it. Anything other than a single ended has a noise penalty . . . but we know a single ended solution is also not ideal.

[Bonsai]

"To add insult to injury, building an ultra low noise current source to feed a LTP is a problem by itself. "

Howland current pump might be the answer here using low noise op-amps.

[mos57]

Great circuit and very well selected components, Syn.

I will try to build a very low noise preamp as this, and i have bought for this purpose some 2SC2545/2SA1083, 2SC2547E/2SA1085E.

My doubt are related how what type of input devices topology: emitter follower or base grounded topology?

Have you compared the two configurations or at least can you tell me because you have choiced the follower circuit for the first stage?
Which advantages?

Thanks in advance, Francesco.

[syn08]

Originally Posted by Bonsai

"Unfortunately, differential stages come with a 3dB penalty in noise."

but does the balanced input stage also not suffer the same fate? Seems we cannot have our cake and eat it. Anything other than a single ended has a noise penalty . . . but we know a single ended solution is also not ideal.

Originally Posted by Bonsai

"To add insult to injury, building an ultra low noise current source to feed a LTP is a problem by itself. "

Howland current pump might be the answer here using low noise op-amps.

Haven't looked closely, but on top of my head I think a balanced configuration will not come with a noise penalty, however I wouldn't know how to apply the headamp feedback loop.

The Howland current pump noise is anything but low, and it's not about the opamp noise, but about the matched resistors network.

Yes, the single ended configuration is not ideal but, from a noise perspective, it's still the best.

[syn08]

Originally Posted by mos57

Great circuit and very well selected components, Syn.

I will try to build a very low noise preamp as this, and i have bought for this purpose some 2SC2545/2SA1083, 2SC2547E/2SA1085E.

My doubt are related how what type of input devices topology: emitter follower or base grounded topology?

Have you compared the two configurations or at least can you tell me because you have choiced the follower circuit for the first stage?
Which advantages?

Thanks in advance, Francesco.

Francesco,

From a noise perspective, it would be about the same. However, you won't be able to get a high impedance input with a common base circuit. If this is not important for you, then certainly the common base circuit has some advantages.

Actually one idea that I am further contemplating is a floating input stage, similar to prof. Leach circuit. I would certainly use common base in such a circuit. The big advantage of a floating configuration is that you can get zero MC cartridge current, even with unmatched bipolars, since the MC cartridge is not referred to ground. It's not that simple, though