|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
63.246.183.22
Q: When is Full-Wave not Full-Wave?A: When it's a voltage doubler.
Explanation...
Below is a schematic of a full-wave voltage doubler:
Note that the output ripple frequency of this supply is 120 Hz. By definition, it's a legitimate full-wave rectifier. However, something is happening internally, a hidden downside to the design. In addition to the 2X increase in current demanded from the power transformer (compared to a standard full-wave or full-wave-wave bridge), there's a degradation in effective filter capacitance that's easy to miss.Looking at the top of C2, there's 75uf of total capacitance (two series 150uF). If each cap has an ESR of 2 ohms, total ESR is 4 ohms. Combining the ESR with XC, the total Z across the circuit should be 18.14 Ohms. Unfortunately, this is an illusion.
Below is a SPICE plot showing output ripple voltage (Blue) and currents through C1 (Green) and C2 (Red). The ripple is indeed 120 hz, but currents through the filter capacitors aren't. Each cap only sees a 60 Hz charging current. This greatly reduces the filtering ability of these components compared to a standard rectifier/filter. Halving the frequency doubles XC. It also doubles ESR. This means the two series caps actually exhibit 8 Ohms ESR and 35.4 Ohms XC. These combine to create a total shunt Z of 36.28 Ohms, exactly twice the previous number.
None of this was apparent to me until just recently. Harman Kardon used this topology frequently in their amplifiers and receivers, and I'm rebuilding several of those. As a result of this analysis, I'll be making substantial changes to add more energy storage and to reduce ripple in all of them.
--------------------------
Buy Chinese. Bury freedom.
Edits: 10/15/16Follow Ups:
The difference between half wave and full wave is 2 diodes for the half and 4 for the full wave
Lawrence
You have hit the only real disadvantage of the full-wave doubler vs. the full-wave bridge, as the ESR is higher at 60 Hz than at 120 (not double, though - the few data sheets I can find say 1.2 - 1.7x). Not a problem with modern caps. Output ripple will be near identical, DC current rating the same for a given transformer size - for the doubler, winding resistance will be 1/4, since the winding will have twice the copper cross-section, and half the length. Either is superior to the full-wave center-tap which can supply about 25% less output from the same size transformer.
FWIW, I say "full wave". What you have, in fact, is 2X 1/2 wave rectifiers wired back to back. However, the liabilities usually associated with 1/2 rectification, like "standing" DC on the trafo and only 1/2 of the waveform being "captured", are not present. The construct can exhibit low copper losses. Combine low copper losses with the cost and limitations of the SS diodes of the early 1960s and it's (IMO) obvious why H/K, Fisher, Marantz, McIntosh ... gravitated towards the "full wave" doubler.
As PSUs are inherently differential, not single ended, that very diode/cap. setup could be used for a bipolar supply and each rail would exhibit a 60 Hz. ripple frequency.
A 100% true full wave multiplier is possible, but it's complex and costly. Look at the last (full wave series-parallel) setup on the linked page. IMO, motor run parts are needed in the connections to the AC feed.
Eli D.
You lost me regarding the copper losses. The doubler causes 2X the current at the transformer's secondary for a given PS output power (vs the usual full wave supply). If the secondary wire gauge isn't upgraded to accommodate this, losses will be higher. I'm with you regarding the diodes, although they did know how to series them back then. I have an RF amplifier with a 1,500 VDC @ 1A supply from the early '60s. It's all solid-state using strings of diodes (four 811As as finals).
Yes, I've looked at a few of the other doubler circuits. They all seem to have some drawback when I SPICE them, usually more difficult to overcome than this one.
--------------------------
Buy Chinese. Bury freedom.
The H/K Cit. 2 has the requisite copper mass in the rectifier winding and does exhibit low copper losses. Remember, higher AWG needed, but fewer turns.Stu Hegeman called for series wired pairs of "top hat" diodes in the "Duece". In their day, those diodes were about the best available and they were expensive.
We take 1000 or greater PIV for granted, these days. Stuff like that did not exist in 1960.
Eli D.
Edits: 10/15/16
I like doublers,especially when the transformer was specially wound for it as in the case of the Mac Mc240 and the Deuce.
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
JOC, how did they make the Kilo-volts needed for transmitters, in the 1940-50s?
Thanks!
The 5R4GYB could handle powers up to a few hundred watts. Mercury vapor rectifiers like the 866 were used up to 1KW or so. Above that, commercial broadcast transmitters used larger mercury vapor tubes or high-vacuum rectifiers. The 8008 MV rectifier is rated at 10KV inverse peak and 5A peak current. The 8020 vacuum rectifier is rated at 40KV inverse and 750 mA peak. I'm thinking a couple of those would work well in my next line stage. :)
--------------------------
Buy Chinese. Bury freedom.
To have KV+ rating? Or did they use a series of, say, 450VDC caps --- in a string?
I have a RF linear amplifier from the '60s with a 1,500V/1A solid-state supply. It's a choke input filter with series 450V electrolytics. I don't know what the high power broadcast transmitters use, I've never worked with them.
Incidentally, I still remember going to see the "artificial lightning" demonstration at the natural science museum in Philadelphia when I was a boy. The demo charged a large bank of capacitors, then let go when there was enough voltage to arc across the gap. During the demo, there was a huge bang, following which a block of wood in the path of the arc literally disappeared. The high-vacuum rectifier they used was about three feet tall, lit up like a bunch of 100W bulbs. Yowee!
--------------------------
Buy Chinese. Bury freedom.
Scary, just thinking about it. Boy, to be a tech in those days!
8^)
Steve
They were using tube rectification in those days and they were good sized rectifiers along with the power supply iron and the modulator iron.
Take a look at this Collins broadcast transmitter manual.
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
Not sure I'm following you, here.
Educate me!8^)
Edits: 10/15/16
Well, it's shown in the SPICE analysis. The blue line indicates the presence of 120 Hz ripple frequency at the top of C2 in the schematic. This is the output of the power supply. What this seems to imply regarding the C1/C2 string - that the caps are filtering 120 Hz and conducting AC currents at that frequency - isn't true. The connection between the two diodes and the caps is such that each cap individually sees 60 Hz, not 120 Hz. Here's how it works:
When the top of secondary L2 goes positive, D1 turns off and D2 turns on. Conduction through D2 charges C2 for a half cycle. When the transformer reverses, D1 turns on and D2 turns off. Now C2 is no longer being charged. Instead, current through D1 and the transformer winding charges C1. This means C1 and C2 each receive charging current over half the 60 Hz waveform. At the top of the stack, one pulse of the 120 Hz wave is created when C2 gets its charge. The next pulse occurs because the charge on the anode of C1 that occurs a half-cycle later causes a voltage rise that's applied to the cathode of C2. That rise causes the anode of C2 to also rise, due to the low impedance of C2 at 60 Hz. So, even though each cap is charging at a 60 Hz rate, the charge currents occur on each cap on alternate half-cycles, and they add in the series stack to create the 120 Hz wave at the top.
--------------------------
Buy Chinese. Bury freedom.
Got it, thanks. So, the residual 60Hz you are seeing is due to the frequency each cap is charging?
I'm not sure what you mean by "residual." Ripple output from the supply is 120 Hz. Yes, it's created by the caps being charged at 60 Hz on alternating half-cycles.
--------------------------
Buy Chinese. Bury freedom.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: