jayme said:
I'm not sure either, supposedly it can handle higher currents than using rectifier tubes alone. I did try it out briefly in my tube amp and found that it was only marginally better than an all-diode bridge, so I removed the "bypass diodes" on the vacuum tubes and ended up with quite a bit less noise and better sound. I'm sold on the hybrid bridge concept, but still hope to move to a tube only rectifier if a suitable power transformer ever turns up at the local surplus store.
Actually, that should replace reason #1 outright. "Fashion" as a reason for using a tube rectifier doesn't mean anything and only smacks of cynicism. Like most DIYers, I've tried both approaches and prefer the sound of tubes. The day I limit my listening to sine waves and square waves, I'll then rely on measurements to predict best performance.
John
I tend to agree with Dave, although I’m not comfortable with why. Even with heroic attempts in shielding and grounding, I sense greater “ease” with a tube rectifier in the case of capacitor input filters. A recent headphone amp project is a case in point. Despite being regulated throughout and carefully grounded, it still sounded better to my ears (inside Sennheiser HD-600s) with an EZ81 instead of FR307 fast-recovery diodes. However, I do find solid-state rectifiers to work as well as tube rectifiers with choke input supplies. Perhaps this is because the much lower peak currents create much lower stored charge in the SS rectifiers. Even so, I still choose fast SS parts there, out of audiophilia nervosa as much as anything else.
And yes, tubes rectifiers do look better too.
And yes, tubes rectifiers do look better too.
Shoog said:
Hi Shoog. Sorry, it's catastrophe day here at work. Different ringing. This is subsonic or low frequency ringing, not ultrasonic. An LC power supply is a sequence of resonant reactive components tuned below 20 Hz (normally) which relies on resistance for damping. What some would consider all ideal coponents - zero DCR iron and no loss capacitors - would ring for seconds on excitation. Try setting all resistances to zero in Duncan's sim software and watch the fun.
Rectifier noise can showw up in AM radio receivers. The usual solution to this in "All American 5" tube radios is to connect an 0.01 at 1KV or so cap across the rectifier diode, and to introduce a small resistance in series with the diode-cap circuit. The cap used should have at least the PIV expected on the diode, and more for safety margin. This gets rid of RF induced buzz on strong AM stations, and hash on weaker signals. This is also done in solid state radios too. Some portable AM and SW radios specify the use of certian wall warts for line operation, as most wall warts don't have the caps across their diodes to avoid the buzz and hash.
I'd think that you'd want a power supply to replicate a stout battery for B+. That sag is not a feature.
Found this: http://www.nervenet.info/HT_delay/htr214xp.html
But hey, if SY says it's ok for <1KV psu systems using SS rectifiers to not have a delay then I guess it's one less circuit to build?
But hey, if SY says it's ok for <1KV psu systems using SS rectifiers to not have a delay then I guess it's one less circuit to build?
Thanks for that. I of course was refering to diode induced ringing, but I wasn't certain of the destinction myself.
One advantage of a switch on delay is that it can eliminate switch on noise as everything is warming/charging up. There so simple to implement I have simply got into the habit of using them.
Shoog
JojoD818 said:
...to wait until tubes got hot, then punch hot tubes by a voltage hammer?
Thanks SY for pointing this out - it can never be too emphasized.
Series impedance of SS diodes while conducting is frequently orders of magnitude smaller than that of a rectifier tube. Consequently, all stral L's and C's of your transformer and capacitors will find a nice almost completely undamped path to resonate through freely, having a long recovery diode just makes it worse because the conducting phase will also have a period where the diode conducts even reverse biassed.
In order to compare, the series impedance has to be rised to the same level as that of the rectifier tube we are comparing with. A simplistic b ut sufficiently accurate way to do this is to pull up the rectifier tube's If vs Vf graph, find out what themaximum ripple current is, and then read of the forward voltage for that current. Divide latter by former, and you get an approximation of rectifier forward 'impedance'. Use this as a series resistor for the SS diode, and then compare. Finally, replace FRED with regular diode keepig the series resistor and compare - you may be surprised!
SY said:
SY,
Good point. No, I didn’t equalize for the equivalent series resistance difference. The EZ81 has roughly 100 to 150 ohms of series resistance where I’m using it. The secondary DCR of the power transformer is about 275 ohms per leg (CT to HV). Remember this is a smaller headphone power amp design. So the total series resistance would have been roughly 285 ohms for the SS case compared to maybe 400 ohms for the tube rectifier case. I suppose that this 1.4:1 ratio in resistances could make an audible difference due to lower peak currents in windings and better resonance damping, but I just wouldn’t have guessed by so much. I offer no better analysis, besides wondering about stored charge in the SS diodes making high-frequency switching transients. Perhaps these transients radiated beyond my careful grounding and layout and were picked up by conductor loops in the audio circuits within the shared chassis. I dunno. The filter, in either case, is a C-R-C: 47u/133R/47u. Low PS resistance and absolute voltage level are not concerns since this raw DC then feeds Harris HIP-5600 regulators chips. These are no longer made, unfortunately. They were like 400 volt versions of the LM317 (lower maximum current rating of course). I think of it as a Maida design in a one chip. One device makes 255 volts for the front end, another makes 170 volts for the output stage (PP 7119s).
Putting the SS diodes back in with series resistors might make a good rainy-day experiment, but I don’t have many free rainy, or even sunny, days lately!
anatech said:Hi SY, ilimzn,
What ????? !
A level playing field? How dare you!
Especially important with DC heater circuits.
Yes, where DCR is especially low. However, I refuse to comment on Counterpoint SA100s again
Brian Beck said:
Certainly a 1:1.4 difference should not be so audible, but we are to an extent still comparing apples to oranges. The distributed DCR of a transformer comes accompanied by a distributed parallel capacitance. Not much, but compared to the rectifier and resistor? Add to that what frequency range we are discussing when it pertains to HF ringing, and the hidden differences may turn out to be much more prominent. In any case, if you have a signifficant voltage drop from rectifier to output of power supply, it is always a good idea to 'move' some of it via resistors, in order to reduce the ripple curent as much as possible. It makes a lot of things so much easyer.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.