Makes not much difference in voltage swing capability in a simple common source amp ?
With higher voltage Schottkys now available would they be a candidate for low level electronic rectifiers, or are the best selected fast / soft recovery diodes the better choice. (Yes i know test them and find out, but I don't have a spectrum analyser - I spent all my money on Beeswax fuses and now my system buzzes - the label on them says "another Eric the Half a Bee product")
You might consider listening tests instead; they are significantly less expensive than a spectrum analyser.
Not really. They are certainly more time and labor intensive if you want any validity. Good spectrum analysis is damn cheap these days.
It seems to me his problem is towards MHz.
Here is my experience with my cascodes:-
http://hifisonix.com/wordpress/wp-c.../Cascode-Oscillation-in-Audio-Amplifiers1.pdf
Here is my experience with my cascodes:-
http://hifisonix.com/wordpress/wp-c.../Cascode-Oscillation-in-Audio-Amplifiers1.pdf
Last edited:
Thanks Mark (and others) for your input on sonic differences of rectifier diodes. We used the Harris (Fairchild) diodes because they fit the space in the SCP-2 power supplies, and they sounded good as well. Parasound uses a different high speed rectifier for several of my designs, but they probably work just as well.
Sometimes, high power supply rejection is just too much a tradeoff to get, against using a superior circuit topology. Unfortunately, folded cascode without a current source feeding it is prone to difficulty with power supply rejection, so I go to all lengths to keep it from getting corrupted by power supply weaknesses, like diode garbage generation.
For my more standard circuits, like my more cost effective power amp designs, I do not demand extra fast rectifiers, as we are pushing pennies around to keep the cost within a range, but with my best stuff, it is mandatory, even with the JC-1 power amp. This, and a number of passive upgrades makes a good amp better (subjectively, I believe) but is hard to measure, and certainly adds to the cost. Yet the reviewers love it, and we have been on the 'A' list for a dozen years with the JC-1, AND we do not advertise in the mags on a regular basis.
A reminder that this thread is originally about the Blowtorch preamplifier, a design with almost no limits in trying to get the best performance, and certainly these power supplies (as well as the Vendetta Research SCP-2 updated) use hi speed soft recovery diodes, BUT they also contain a number of folded cascode circuits that make power supply rejection and the gain stage level, impractical, or it was at the time of their design decades ago, so attention to the power supply noise can be important.
Sometimes, high power supply rejection is just too much a tradeoff to get, against using a superior circuit topology. Unfortunately, folded cascode without a current source feeding it is prone to difficulty with power supply rejection, so I go to all lengths to keep it from getting corrupted by power supply weaknesses, like diode garbage generation.
For my more standard circuits, like my more cost effective power amp designs, I do not demand extra fast rectifiers, as we are pushing pennies around to keep the cost within a range, but with my best stuff, it is mandatory, even with the JC-1 power amp. This, and a number of passive upgrades makes a good amp better (subjectively, I believe) but is hard to measure, and certainly adds to the cost. Yet the reviewers love it, and we have been on the 'A' list for a dozen years with the JC-1, AND we do not advertise in the mags on a regular basis.
A reminder that this thread is originally about the Blowtorch preamplifier, a design with almost no limits in trying to get the best performance, and certainly these power supplies (as well as the Vendetta Research SCP-2 updated) use hi speed soft recovery diodes, BUT they also contain a number of folded cascode circuits that make power supply rejection and the gain stage level, impractical, or it was at the time of their design decades ago, so attention to the power supply noise can be important.
I like the suppositions about hard measurable parameters like ESR, capacitance, etc. but "yeah oh the scope showed nothing". Schottky diodes faster than the music.🙄
Problem: folded cascode circuits without current source drive, are very sensitive to power supply imperfections.
Solution 1: add current source drive to the folded cascode circuits.
Solution 2: make the power supply perfect.
What's wrong with solution 1?
Solution 1: add current source drive to the folded cascode circuits.
Solution 2: make the power supply perfect.
What's wrong with solution 1?
Making a low noise current source can be a problem, because active current sources magnify their intrinsic noise. It is not always practical to make a very linear and low noise gain current source. Remember Gm in the active devices.
Solution one leaves noisy nodes in closer proximity to the circuitry, but this is not necessarily a significant problem.
Also, upgrading a folded cascode alone doesn't improve supply noise rejection in other parts of the amp. For a frontend, one transistor filtering the rails might be more efficient than 6 transistors cascoding everything individually, which tends to lead to very high transistor count amps.
Also, upgrading a folded cascode alone doesn't improve supply noise rejection in other parts of the amp. For a frontend, one transistor filtering the rails might be more efficient than 6 transistors cascoding everything individually, which tends to lead to very high transistor count amps.
Last edited:
The issue with diodes that cause switch-off ringing is not that they will pump garbage into the circuit on the rails (see OS's comments - I've also never seen HF on the rails of my PSU's using standard rectifiers). The problem is they cause HF to be radiated in the wiring between the reservoir caps and the transformer secondary wiring. This can then possibly be picked up by the amplifier circuitry and especially so if we are talking small signal stuff.
However, the cure in all these cases it to snubber effectively and to minimize loop areas to reduce radiation - for the 50/60 Hz stuff and any remaining HF stuff. So, it not magic rectifiers, but sensible EMI engineering that's required. I put my snubbers directly across the ~ pins on the rectifier and usually use 1n + 1k - probably best to check individual designs though.
Fast diodes with soft recovery are a must in HF SMPSU's since snubbers can result in considerable power dissipation = lower efficiency. I just don't think they are a good design choice for standard mains transformer based PSU's.
🙂
Yes, one of the few cases I needed snubbers was when some bog-standard 1N400X parts were in proximity to a moderately-low-level circuit, and there was an audible buzz from the demodulation of the reverse recovery spiking. Sometimes one sees only a capacitor across the given diode, which in itself is not very dissipative, but pushes the resonance down in frequency so that it radiates with much lower efficiency. However then if the line noise is really terrible this can couple more of it across the off diode, so there is a tradeoff.
It is in the current, Bonsai, so you need a current probe (or its equivalent) to see it easily.
The LA article mentions several injection mechanisms including those in #81917 & 81918; see 2nd paragraph on 1st page.
In DIY gear where the builder is not "pushing pennies around" like John Curl's low priced commercial products, it strikes me as ridiculous and even foolish to accept any amount of ringing whatsoever. The complete and total cure is well known and very low cost. Me personally, I use soft recovery diodes and I also overdamp the transformer secondary RLC circuit to prevent ringing. Costs a couplea three bucks more. I can afford it.
Ringing is an unsightly wart; the surgery to remove it is low risk, low cost, and 100% effective.
In DIY gear where the builder is not "pushing pennies around" like John Curl's low priced commercial products, it strikes me as ridiculous and even foolish to accept any amount of ringing whatsoever. The complete and total cure is well known and very low cost. Me personally, I use soft recovery diodes and I also overdamp the transformer secondary RLC circuit to prevent ringing. Costs a couplea three bucks more. I can afford it.
Ringing is an unsightly wart; the surgery to remove it is low risk, low cost, and 100% effective.
- Status
- Not open for further replies.