• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Choke ratings and resistor values in choke input supply?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Better?

01B1FC98-D5C4-450E-A71A-D00B495C982D-37165-000019C37827DBE7_zps4ba67876.jpg
 
Both damper cathodes and both damper heaters should be at the same potential. Then, no issue of heater to cathode potential limit can surface.

Protection against inductive kick back spikes is needed, when SS diodes are mixed with choke I/P filtration. I don't see "sand" in this B+ PSU. :yes:

The voltage equalizing resistors shown will slowly bleed the filter caps. down.

If the B+ supply is charged up, so is the bias supply. Negative voltage on the O/P tubes' control grids provides electrostatic protection against cathode stripping. ;)

Speaking of the bias supply, mention was made of having both a "free" 5 VAC winding and a free 6.3 VAC winding. Phase those windings up and wire them in series. Fewer stages of voltage multiplication will be needed, to obtain a satisfactory bias supply.

If a "fudge factor" cap. proves necessary, make sure you use a part rated for several thousand WVDC. The 0.47 μF. limit for the "fudge factor" part is to ensure that a cap. I/P filtered regime is not entered.
 

Attachments

  • Damper Rectified PSU.jpg
    Damper Rectified PSU.jpg
    56.8 KB · Views: 193
Last edited:
I re-ran the sim with your current schematic but without the bleeders or the small input cap (both of which are unnecessary). I had to guesstimate transformer primary DCR, which makes a big difference (I used 5 ohms), and just for grins I added parasitic capacitances to the chokes. Rectifiers are 6D22S, since I don't have a model for the 6CJ3. They are very similar tubes, so the voltage will be accurate within a couple volts.

With this configuration you are looking at 503V at idle (110mA) with almost zero ripple. At 350mA load the output voltage drops to 427V. Regulation is 18%. The loss of regulation over my earlier sims is due to the DCR of the trans primary and the DCR of the second choke. If you measure the DCR of the trans primary I can get even better accuracy.

One area of concern is choke saturation on startup due to the large final cap size. With the choke parameters I have entered, the rectifiers must endure a peak current of 1.2A about 60ms after startup, with current over 1A lasting for several cycles before tapering off to idle current after about 250ms. If the chokes saturate the current will be much higher, potentially limited only by the DCR of the trans and chokes. This is something to look at once the supply is built. Damper diodes do not have a hot switching transient current rating on their data sheets like tubes designed for power rectifier service. They do have a "peak" current rating (2100mA in the case of the 6CJ3), which is called out as a "steady state peak current". This combined with the rugged cathode means the damper diode should handle a couple amps for 200ms. The only caveat is to make sure the cathode is hot before switching on the plate current. A lot of folks suggest using the slow warm up of dampers as a slow-start feature. This is really bad practice. Yes, they are tough, but they're not indestructable. If you want a slow start, use a step-start arrangement.

Oh, and on the subject of the bleeders. If you don't need 'em, don't use 'em. They are a waste of power (although they can improve regulation). If you are concerned about over-voltage, use an OVP circuit.
 
Oh, and on the subject of the bleeders. If you don't need 'em, don't use 'em. They are a waste of power (although they can improve regulation). If you are concerned about over-voltage, use an OVP circuit.

I agree. With the exception of "if" in regard to concern. I can think of two circumstances where the voltage will soar to over 1kv without over voltage protection under common use and I don't even have any rigorous electrical training. Just a lot of hands on one-on-one bench experience as I was growing up.

Eli mentions the use of bleeders in choke input filtration as an important one.


Over-voltage protection of some kind with choke input power supplies is important. Bleeders are easy so that is why I suggest them.

I think keeping voltage where it is expected and in the amount it is expected is the only concern. Once that is taken care of, everything else can proceed and we can enjoy arguments of what is hifi etc etc.

I'm sure there is some over-voltage protection in the works and I'm just jumping the gun here. I can't imagine anyone here would recommend there would be none. -Fred
 
Having the voltage "soar to over 1kv" would require a series of extremely unlikely events. Under normal conditions, loss of load on the filter would have the voltage reach 850V across the filter caps. If each cap is rated for 450V, there is no issue. The supply is essentially "passively safe". There is also the rest of the amp circuitry to consider. A 20mA draw from a series divider, for example, will hold the voltage down to 620V. There would be no need to burn another 20 watts in a completely unnecessary bleeder. That is why I said "if"
 
Last edited:
Hi,

Thanks for doing these Sims!! The primary DCR of my transformer is 1.2ohms

I plan to drop the cap size on that second cap to around 100-150uF.

Eli suggested the snubber to tame input peaks and take a small amount of "impact" load off L1 in case it is not quite up to L1 duty even though I confirmed with Hammond that it is, in fact, up to the task.

I will not have a way to energize the cathodes before the plates on the damper diodes.

One question I do have regarding bias supply and the Triad I'm going to use for the filaments of the 6CJ3.

Since it has four isolated windings, can I power the filaments off one complimentary pri/sec, and then feed the other complimentary pri/sec in "reverse" off a spare 6.3V winding to get my bias voltage?
 
What I want to know is would you suggest over voltage protection? -Fred

As I already mentioned, as long as the caps have sufficient WV no OVP is strictly necessary. If OVP is needed for some other reason, I prefer an active OVP circuit that doesn't needlessly burn extra power.

You could contemplate putting 2x series 1N4007 in front of each vacuum diode for simple added protection, and something like a 660VDC MOV across the choke (2x 330VDC in series - probably only need to be 7mm discs), as well as a damping high kohm resistor (again say 2x series 2W resistors rated at say 350VAC each).

What do the series silicon diodes accomplish? The PIV rating of the damper diodes already far exceeds that of the SS diodes and any peaks generated by the supply.

Hi,

Thanks for doing these Sims!! The primary DCR of my transformer is 1.2ohms

That actually makes a substantial difference due to the primary current. Note that I am not taking into account the primary voltage drop due to other secondary loads such as heaters, but this should be reasonable. With that primary DCR you are looking at 513V at 110mA and 459V at 350mA. Regulation is about 12%, which is a significant improvement vs. my previous assumption of 5R primary DCR.

I plan to drop the cap size on that second cap to around 100-150uF.
The 2 470's in series are fine as long as there isnt a huge current spike at startup from choke saturation. This will need to be determined after the supply is built.

Eli suggested the snubber to tame input peaks and take a small amount of "impact" load off L1 in case it is not quite up to L1 duty even though I confirmed with Hammond that it is, in fact, up to the task.
Under normal conditions the voltage across, and current through, the choke is quite reasonable. Under abnormal conditions voltage spikes can arise at the junction of the choke and rectifiers, but they are well within the rating of both components.

I will not have a way to energize the cathodes before the plates on the damper diodes.
You can put a suitable relay in the HV leads, switched by a standby switch. Otherwise, give it a try without. My concern is arcing between the cathode and plate. If it doesn't happen, your good. If it does, you will have to let the cathodes warm up before hitting them with HV.

One question I do have regarding bias supply and the Triad I'm going to use for the filaments of the 6CJ3.

Since it has four isolated windings, can I power the filaments off one complimentary pri/sec, and then feed the other complimentary pri/sec in "reverse" off a spare 6.3V winding to get my bias voltage?

No.

I can get 22, 56, 68, 120, & 220uF 500V caps to series for 1KV working voltage. I know using a fairly small(er) cap as the first cap after the choke is best. So, maybe 2X 56uF for 23uF @ 1KV there, but what is the best size for after L2?

Thanks!

Final cap size is limited by transient current during switch-on. In a choke-input supply, this will be the point where the choke saturates and causes a large turn on surge. You will need to determine this empirically, since I have no way of knowing when your choke will saturate.
 
Last edited:
Just for more grins, I figured I would post some pretty pics.

The last image is the schematic I am simulating. When initially finding the steady state output voltage I short the load resistor and set the current source for the draw I want to simulate (110mA in this case). Once the voltage is found, I use Ohms law to calculate the load resistance, plug that value into RLoad, and short the current source. Then I re-run the simulation. Using a resistive load rather than a current source allows more accurate simulation of startup conditions.

The transformer and chokes all have the series DCR plugged in. The chokes also have parallel capacitance added. The 6D22S's are fairly accurate with respect to load behavior and capacitances, but not saturation. Since the currents are well within the max ratings this does not affect the results.

The first image is output voltage at the load and current through L1. You can see both the voltage overshoot on startup as well as the transient start-up current.

The second image shows voltage at the input of L1 (green) and output of L1 (blue). Just after 200ms the current through the choke falls below the critical level, which causes transient voltage spikes on the input side of the choke. By 350ms the current rises above critical, and the transients disappear. Vacuum rectifiers have no problems with these spikes, but SS diodes may not be so forgiving unless their PIV rating is well in excess of that needed for steady-state operation. RC snubbers can be used to dampen the transients if necessary. The third image shows a closeup of the transient spikes.

Images 4 and 5 show current through the rectifiers and the transformer primary respectively. Images 6 and 7 are close-ups of images 4 and 5, showing the current waveforms under steady 110mA output conditions. Peak current through the rectifiers is only 160mA, and peak line current is around 800mA. Also note the minimal difference between peak and average current, showing good utilization of the transformer, rectifier, and line.
 

Attachments

  • Screenshot from 2013-09-08 03:32:52.png
    Screenshot from 2013-09-08 03:32:52.png
    9.7 KB · Views: 176
  • Screenshot from 2013-09-08 03:38:04.png
    Screenshot from 2013-09-08 03:38:04.png
    10.2 KB · Views: 167
  • Screenshot from 2013-09-08 03:50:13.png
    Screenshot from 2013-09-08 03:50:13.png
    10.5 KB · Views: 172
  • Screenshot from 2013-09-08 03:56:43.png
    Screenshot from 2013-09-08 03:56:43.png
    9.1 KB · Views: 175
  • Screenshot from 2013-09-08 03:57:16.png
    Screenshot from 2013-09-08 03:57:16.png
    7.5 KB · Views: 171
  • Screenshot from 2013-09-08 04:08:57.png
    Screenshot from 2013-09-08 04:08:57.png
    8 KB · Views: 46
  • Screenshot from 2013-09-08 04:08:16.png
    Screenshot from 2013-09-08 04:08:16.png
    9 KB · Views: 47
  • Screenshot from 2013-09-08 03:12:59.png
    Screenshot from 2013-09-08 03:12:59.png
    10.6 KB · Views: 69
Last edited:
I shorted L1 and L2 and re-ran the sim to show the differences in current waveforms and magnitude with a cap input supply. The first image is current through the rectifiers. The second is a close-up of the first under steady state operation at 110mA draw. Images 3 and 4 are the primary current under the same conditions. They clearly show the poor peak-to-average current of a cap input filter, poor power factor, and overall poor utilization. If the transformer core saturates on startup the waveform gets even uglier. The high peak currents cause increased winding and core loss in the trans, making it run hotter (and in higher power apps requiring a larger transformer). Finally, the resulting ripple contains a good deal of harmonic content, whereas the ripple in a choke-input supply is a sine wave of twice mains frequency with almost no harmonic content. Really, the only advantages a cap input supply has is higher voltage (at the cost of worse regulation) for a given secondary RMS voltage, and cost.
 

Attachments

  • Screenshot from 2013-09-08 04:35:36.png
    Screenshot from 2013-09-08 04:35:36.png
    6.5 KB · Views: 75
  • Screenshot from 2013-09-08 04:36:02.png
    Screenshot from 2013-09-08 04:36:02.png
    7.7 KB · Views: 26
  • Screenshot from 2013-09-08 04:36:20.png
    Screenshot from 2013-09-08 04:36:20.png
    5.9 KB · Views: 41
  • Screenshot from 2013-09-08 04:36:48.png
    Screenshot from 2013-09-08 04:36:48.png
    6.9 KB · Views: 35
Last edited:
http://www.turneraudio.com.au/audiofilt ... page2.html

I am looking at this PS site from Patrick Turner (the traditional choke input), and noticed that the 6K bleeders are 30W! Do these really have to be 30W?

The other question is that the second choke (L2 in the image) I have is only rated at 715 WVDC. My supply transformer is 1100VCT. Would a current limiter help the startup spikes? Is it OK to use the 715V choke in the second position?

Thanks!

Blair

Well, with no schematic, I can see this "iron set" is geared towards being a tube rectifier. The transformers and inductors are insulated tested hundreds of volts higher than the working voltage. So 50V should not be a big deal. As for the 30W resistor, depends on the drop across if it needs to be that high of a wattage.
But back in the day, the resistor's insulators had lower working voltage and a 30W comp resistor had a working voltage near 1000V. In any case, a mills 7.5k MRC50 will work nice if you got the real estate and if you are going for a solid state. There 12W would work if it was just the factor of working voltage. But physical test measurements would conclude that. Some people will put in a big resistor in because it doesn't heat up as much as the later.



6k -12K is really low for a bleed. you must be talking about the resistors in a rc filter network.


BTW, your schematic link does the "404 error" .
 
What do the series silicon diodes accomplish? The PIV rating of the damper diodes already far exceeds that of the SS diodes and any peaks generated by the supply.
Part failures and poor connections are uncommon, but sometimes cheap insurance (by adding protective devices) can avoid stresses on other parts, especially where large inductors are concerned. Damper diodes can fail, but I'm uncertain if anode-cathode leakage is likely.
 
Part failures and poor connections are uncommon, but sometimes cheap insurance (by adding protective devices) can avoid stresses on other parts, especially where large inductors are concerned. Damper diodes can fail, but I'm uncertain if anode-cathode leakage is likely.

Vacuum rectifiers, including damper diodes, are quite reliable. Dampers are actually much tougher than common indirectly heated rectifiers like the 5AR4. A-C leakage is generally nil, particularly in the case of dampers with cathode cap connections (which are the most common types). Remember, these tubes saw brutal service in TV sets where they were exposed to repetitive HV spikes of 3-5KV along with high current pulses. They don't need silicon diodes to "protect" them.

I should also point out that choke input supplies reduce stress on the other supply components. It is in fact cap input supplies that are brutal to both the rectifiers and the input caps.

The only real failure that can endanger the supply is one where one rectifier shorts while the other fails open. In that case, AC will be applied to the choke input. The choke will tolerate this, but the following electrolytic caps may not. AC across the first cap bank would be about 30v peak, which 450V electrolytics may or may not be happy about. In this case, a reverse biased 1KV diode across the caps would clamp the negative going pulses. Current would be low, perhaps a couple hundred mA or so, so the fuse would never blow. The amp would simply not work until the rectifiers were replaced, since there would only be about 15V available at the output of the supply. Keep in mind that such a failure is extremely unlikely. It would be like winning the lottery. Twice. If the caps following L1 happen to be film or oil types rather than electrolytics, then even that single diode would be unnecessary.

Back in the day no silicon diodes were used anywhere in the power supplies (because they didn't exist), and millions of radios, TV's, amplifiers, high power RF amps, etc. all worked fine with tube rectifiers. In my opinion, if someone feels that vacuum rectifiers need series silicon diodes to protect them, then they may as well dispense with the vacuum rectifiers all together.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.