Exactly, and the 337 is particularly bad, see the measurements made by Walt Jung in his article on current sources
From a performance perspective there are also simpler phono stages that are "already doing great", but why compromise when you can do better? 2 transistors and a handful of resistors and caps is not that complex.
Are you aware that JC is using the same 'horrible' 337 as preregulator in his Vendetta?
The parallel regulator has 150dB loop gain and 2.6MHz unity loop gain. This makes the current source (be it 337 or a simple resistor) impact practically negligible.
There is a problem with this power supply, far from obvious and not related to the AC performances or anything that Spice can reveal. Hopefully it will not require HPS4.1
Be back as soon as I have answers.
Onlly as a preregulator. I guess you are one of those people who follow JC in everything he says or does and not ever challenging - I am afraid you can't be helped then
Probably right, except that LM337 has a low roll-off, but maybe I am nitpicking again. FWIW, if I cascade two even half-decent regulators in ltspice I get extremely high rejection figures.
P.S. What are those other power supply issues you are talking about?
I don't understand the use of this tradional (and for me bad) kind of power supply stabilizing. Have a look to the NAD RIAA head amp "PP-2" (PP2, see pdf attachement).
Basicly there are only one LM 317 as current source. The value of current must be adjust so, that one half runs through the concerning amplifier stage and the other to the two serial connected shunts (respective zener diodes). For better filtering I use additional one gyrator (electronic inductor) and one real inductor (aprox 0,1mH) in serial between the pos. connection of rectifier/first capacitor and the input of current source.
The main advantage is the use of very big values of capacitances in parallel of the shunts/zeners very close to the actually gain stages without problems while switch off (don't forget serial diode behind the current source). If I have capacitances in the "Farad" aera like the new wima supercap about
WIMA SuperCap MC Doppelschicht-Kondensatoren-Module
the disadvance could be, that I must wait a long time before music comes (perhaps one of you can this check out).
Basicly there are only one LM 317 as current source. The value of current must be adjust so, that one half runs through the concerning amplifier stage and the other to the two serial connected shunts (respective zener diodes). For better filtering I use additional one gyrator (electronic inductor) and one real inductor (aprox 0,1mH) in serial between the pos. connection of rectifier/first capacitor and the input of current source.
The main advantage is the use of very big values of capacitances in parallel of the shunts/zeners very close to the actually gain stages without problems while switch off (don't forget serial diode behind the current source). If I have capacitances in the "Farad" aera like the new wima supercap about
WIMA SuperCap MC Doppelschicht-Kondensatoren-Module
the disadvance could be, that I must wait a long time before music comes (perhaps one of you can this check out).
This power supply would be a disaster for my RIAA preamp. It's also as bad as you can get for any preamp worth more than $50.
Coincidentally, the NAD PP's (powered from a wall wart) are the worst RIAA preamps I've ever heard. There's a testimony about on my web site, that's where and why I started building my own.
You misunderstand my coment. I don't recommend a clone of the NAD PP2 power supply. I recommend the basic circuit regulator topology (please read my previous post exactly)
I don't think so. This basic circuit regulator topology is as bad as you can get, for anything but a $50 preamp. Otherwise said, it's a piece of junk.
SYN08
When you have finished this exercise it will be great to hear your listening impressions of it's sound compared to that which it supersedes; and with special regard areas of improvement.
Edit:
I have just checked your website and found that you use a high output moving coil cartridge. Can you also when listening to the unit test with a low resistance, low output cartridge? Many of us have such cartridges and your design, being based on low noise, should perform with these more difficult cartridges.
When you have finished this exercise it will be great to hear your listening impressions of it's sound compared to that which it supersedes; and with special regard areas of improvement.
Edit:
I have just checked your website and found that you use a high output moving coil cartridge. Can you also when listening to the unit test with a low resistance, low output cartridge? Many of us have such cartridges and your design, being based on low noise, should perform with these more difficult cartridges.
Last edited:
I guessed that was the problem.
Is it only the power supply, or is it the same problem with all the heat sinks in the mid row?
The parallel regulator and the current source. They need to be in the proximity of the low noise input stage (running at high currents, to optimize the noise performance following Ic=Vt*SQRT(Beta)/Rsource). Dissipation (about 4W) is high enough to raise the temperature so that any noise benefit is slowly (in time) vanishing. It ends up into 0.5nV/rtHz and it's clearly audible in my horns.
HPS 4.1 is on the way, PCB is due next week. Jung regulators (far away from the low noise stage) and local cap multipliers, exactly as in HPS3.1. The local cap multiplier dissipate about 1W, that's ok. The other option would be to lay out away the parallel regulator and use force/sense. The overall PCB layout will then become a gamble.
I somehow saw it coming, but I said it's worth a shot, at least to estimate the AC performance (can't really breadboard something like this, with 160dB open loop gain and the open loop unity gain frequency in the MHz range). AC performance is absolutely superb, better than HPS3.1.
Another excellent news is that the input bias current can be trimmed down to under 100nA, which makes a high input impedance impedance possible. With a symmetrical layout (input stage devices approximately on an isotherm re: the OPA552 and the local regulators), it's perfectly stable with the temperature. I used the ro-theta web site online simulator to find out the approximative isotherms.
Listening tests Brianco? Yeah, will do, but I wouldn't rely on my subjective opinions. I'd rather let others do such, build it yourself, and give it a shot. So far, I have excellent feedback from the HPS 3.1 builders.
Not anymore, that was over a year ago. As previously mentioned, I'm currently using a Benz Micro SL Wood, 0.4mV and 12ohm. I have a few others (including a 0.2mV Ortofon), but this is my favorite.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.