If you have a centre-tapped transformer that gives the correct dc voltage with full-wave connexion, then a hybrid-bridge will not improve the power supply - and may well make it worse from all the recovery (switch-OFF) current pulses in the semiconductor parts.
if you want double the voltage though - just leave the CT unconnected, and put a hybrid bridge across the two outer terminals of the secondary.
Or you could use choke-input filtering to bring the volts back down a bit, and that would improve the sound greatly, compared to cap-input filtering.
if you want double the voltage though - just leave the CT unconnected, and put a hybrid bridge across the two outer terminals of the secondary.
Or you could use choke-input filtering to bring the volts back down a bit, and that would improve the sound greatly, compared to cap-input filtering.
Obviously no schematic so not exactly "seen anyone do this" but Leben uses a setup like this on the CS-660P...ie SS diodes followed by a damper, but from what I understand mostly for a nice soft warm up. Anyways it doesn't seem to hurt sound quality since it has gotten praise from reviewers, although reviews may or may not need to be taken with a grain of salt.
http://www.lebenhifi.com/products/cs660p.html
Cheers
James
tubesmuggler said:
As other posters have indicated, you don't use a bridge with a CT trafo, unless you are interested in rectifying the entire winding.
You could use what I refer to as a "cockeyed" bridge. Prepare 2 sets of parallel wired UF4007 diodes and 10 nF. HIGH WVDC capacitors. Connect the cathode end of each assembly to an end of the rectifier winding. Ground the anode ends of the assemblies. Connect the plate of the damper to the CT of the rectifier winding. Take the "raw" B+ from the damper's cathode. In this configuration, the damper is not rectifying. Instead, it is blocking any SS diode switching noise the snubber caps. fail to kill and provides a soft rail start.
If a vacuum diode is in series with a solid state diode, I can't imagine how you would get reverse recovery current flow. The vacuum diode would have to conduct in reverse as well, which it doesn't. The only current flow would be from any charge stored in the solid state diode and it seems that this would be like dumping a glass of water into the ocean.
I've never taken any measurements on this or anything, just trying to reason it out.
This is true at first blush. But when you find a (usually time-domain) diagram of the recovery pulse, you see a current peak (magnitude roughly same as forward current peak) with a rise time of less than 10ns, and a decay downward to the reverse recovery time specified by the diode (eg 75ns for the UF4007 - everyone's favourite hybrid bridge component).
Take a fourier transform of the pulse and you'd find spectrum all the way to the GHz region (driven by that rising edge speed) with lumps and bumps through the tens & hundreds of MHz, and finally cross-products all the way down into the ultrasonic and AUDIO range, driven in part by the conduction-angle (ie ON-pulse width) of the diode - which varies with the output power of the load.
Looking at a vacuum damper diode data sheet (6CJ3) the capacitance across the anode-cathode is of the order of 10pF. So the GHz components of the recovery pulse will only see (the order of) 10 ohms in crossing your damper.
These pulses carry over with them the audio-modulated cross-products of spectrum generated by the varying load - right into your amplifier! Your (leaded) B+ filter caps will not work on these at all, because their effectiveness reaches only to low MHz even for good film types MKP, MKT.
Even without the damper capacitance vector, the high frequency mush will get into your amplifier if the rectifier wiring is any length: the quarter-wave antenna for 900MHz radio is about 3.5 inches.
The cure? Snubbing capacitors right on the rectifier/secondary winding to cover all the frequencies they can handle, plus VERY short wiring to avoid transmitting anything else.
Of course, generating the least amount of recovery pulse is best. Cree Silicon-Carbide diodes, anyone?
For anyone with an extreme inclination, put the transformer/rectifier in a cage and feed the wiring, netz & all, via microwave feedthrough caps and ferrites with GHz ratings and 500V durability.
There is a tiny amount of stored charge in a vacuum diode - the amount of charge actually in flight between cathode space charge and anode at the moment the anode potential reverses. How quickly this returns to the cathode would depend on the reverse voltage. In a normal rectifier circuit this charge will have greatly reduced as the AC voltage swings down to zero and the current drops so nothing to worry about.
I guess you could estimate what this charge will be, from anode current, anode voltage and anode-cathode distance. It will be small!
I guess you could estimate what this charge will be, from anode current, anode voltage and anode-cathode distance. It will be small!
With electron transit times in the 1 nanosecond range, the charge in the electron stream will be well below a nanocoulumb (the charge a 1 nF cap will hold at 1V, or a 10 pF cap at 10V). This charge will reverse direction and dissipate in the cathode) over a relatively long period (hundreds of microseconds) as the AC voltage reverses.
Last edited:
Rod, are you referring to the typical reverse recovery waveform in a datasheet or switchmode document? Those measurements typically show the setup as applying a high level of pre-existing on-state current, and a very high level of dI/dt as compared to what would exist in a mains rectified situation.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.