Source for 2-4 Uf Capacitor 3000V for Modulator Projector

[kx9dk]

I'm building an outboard modulator using a PA amplifier with audio xformer on output connected to 40h choke through 2-4 uf capacitor (modified Heising circuit from AM Window). Got all the parts I need, but having trouble finding the 2-4 uf cap 3000v. Any ideas where these can be found??? Will be using to modulate homebrew single 807 transmitter feeding a single 3-500Z amplifier.

Thanks
Dave KX9DK

[KE6DF]

Epay usually has them:

http://www.ebay.com/itm/GENERAL-ELECTRIC-3000V-D-C-4UF-OIL-CAPACITORS-PAIR-/220957212696?pt=Vintage_Electronics_R2&hash=item3372138018

http://www.ebay.com/itm/1-Westinghouse-2-MF-3000-V-DC-PIO-Capacitor-Audio-Radio-Amplifier-NOS-/180823040764?pt=Vintage_Electronics_R2&hash=item2a19e4a2fc

Here is a more reasonable price

http://www.ebay.com/itm/GE-4-mfd-3000-WVDC-oil-filled-capacitor-Type-CG-48270-/160714097222?pt=Vintage_Electronics_R2&hash=item256b4e8246

[Opcom]

edit 3/15/2012 - - I offered the OP some caps, in any case after a few days several others e-mailed me about these caps and asked to take them! how about that. -so removing the offer to prevent confusion later.

[Opcom]

edit 3/15/2012 - to avoid confusion - most of the proffered caps are gone, so the offer's over anyway as the OP didn't need/want them.

[k4kyv]

Why do you need a 3 kv one for modulating a single 807?  Around 1 kv should do the trick.  A plate modulated 807 won't stand much more than about 750 volts on the plate.

I run 2000 volts on the plates of the HF-300s, and the capacitor is 4 mfd @ 3kv. I have been using that same capacitor since I first built the rig in about 1971, and it was ancient even then.

[IN3IEX]

In fact if the supply voltage of the output stage of the audio PA is the same voltage used for the 807, the DC voltage on the capacitor will be nearly zero...
Indeed, there will be some voltage because the DC resistance of the transformer winding and the DC resistance of the Heising inductor are probably different, and different are the currents.. maybe less than a hundred volts of DC on the capacitor. Just apply the voltages simultaneously. Use a high capacitance, this will avoid higher voltages at low modulation frequencies. About 16uF should be ok. 
Maybe use a 1uF 500 V for testing, than as soon as you discover the polarity of the differential DC voltage drop use an electrolytic capacitor (PC power supply 330 uF 450V or what you have at hand) with such a high capacitance there will be almost no AC voltage on the capacitor (therefore high capacitance to avoid any AC voltage added to the DC voltage). Just another approach to computing capacitance...

[The Slab Bacon]

The 4@5Kv is a pretty kool looking cap! Rather odd shaped can.

[kb3ouk]

here's how he came up with the 3000 volts.

http://amwindow.org/tech/htm/obmod.htm

4) A 2-4 uF cap with a voltage rating of 3-4 times the B+ on the final.

750 volts times 4 equals 3000 volts

[Steve - K4HX]

Use a high capacitance, this will avoid higher voltages at low modulation frequencies. About 16uF should be ok.

Do not do this. The transient/impulse response of the circuit will be horrible resulting in excessive modulation overshoot. There was a reason the blocking caps in BC transmitters usually were 2 uF.

[k4kyv]

here's how he came up with the 3000 volts.

http://amwindow.org/tech/htm/obmod.htm

4) A 2-4 uF cap with a voltage rating of 3-4 times the B+ on the final.

750 volts times 4 equals 3000 volts

I'm not sure where that 3-4 X figure comes from, but I've never observed it, nor have I ever seen it used in a BC transmitter.  That would mean, for a plate voltage of 2500v for example, you would need a 10 kv capacitor.  Most transmitters use something like 1.5 X plate voltage for adequate safety margin, comparable to the voltage rating you would use for the filter capacitors in the +HV plate supply.

Also, there are several alternatives for connecting the capacitor.  I would not use the one shown in that diagram with a regular modulation transformer, since that would put the secondary winding at ground potential while the primary is at +HV, resulting in excessive stress on the mod xfmr insulation.  But the diagram shown would be the preferred connection for a backwards connected output transformer, since the whole thing can be run at ground potential.

I have tried higher values of coupling cap and never noticed any overshoot, but that was not something my attention was on at the time. OTOH, excessive coupling capacitance can produce a transient problem, due to the prolonged charge-up time and inrush current when wired so that the final amp DC voltage appears across the capacitor. That is the reason I always thought the 1-2 mfd cap, but rarely something like 16 mfd, was used. Plus, for economy reasons, since 1-2 mfd is adequate for the job at usual modulation impedances, and HV oil caps were expensive, particularly in the WW2 and pre-war era. I would think any overshoot problem would depend on the wiring configuration, inductance of the reactor and that of the modulation transformer.

If the bottom end of mod transformer secondary is returned to the +HV lead from the final amp power supply instead of to ground, no DC appears across the capacitor, and the only voltage rating required would be to handle any audio transients that might appear across the capacitor.  I would still use a voltage rating at least the same as the DC plate voltage.

One school of thought is the brute-force method: use a large coupling capacitor and as high inductance in the mod reactor as practicable, so that the capacitor looks like a dead short throughout the audio range, while the reactor looks like an open circuit, and the mod transformer is essentially working into a resistive load (the modulated PA) while the capacitor and reactor appear invisible to the mod transformer as it works into the PA load.

The other school of thought is to use values of capacitance and inductance to make the reactor and capacitor act like a high pass filter, with a roll-off frequency chosen to be near the lowest audio frequency to be transmitted. I am a little leery of that method; I would think it would result in excessive phase shift at the low end of the audio range. I believe it was Timtron who told me about a Gates 5 kw transmitter he worked on that used that method in the design.  It had an unusually low-inductance reactor for the modulating impedance, and he said it had a hump in the audio response somewhere in the vicinity of 50~.

In their old catalogues, United Transformer Co. (UTC) recommends that all the audio transformers in the system have as close as possible to a flat frequency response over the range of at least one octave above and one octave below the actual frequency range the amplifier is expected to handle.  The purpose of this is to keep phase-shift distortion to a minimum throughout the usable range. That means if the transmitter is expected to carry a high-quality voice frequency range of 100-5000~, you would design the modulator to have a measured flat response of 50-10,000~, which is typical of what most tube-type BC transmitters with transformer coupling actually have.  Even if the targeted frequency response is "communications quality" 300-3000~, you would want the modulator to be capable of flat response from 150~ to 6000~.

[kx9dk]

Wow!,

Thanks for all the responses and things to think about. Yes, that is the circuit I got the idea from.  http://amwindow.org/tech/htm/obmod.htm

I have an old PA amplifier and a 100 watt audio transformer to be connected backwards that will go through a cap to the top of a 40h reactor. Yes, B+ through the choke to the 807 is only about 600 volts. I did check out epay (thanks for the links WB6IYH). Got an accepted offer on a 4uh, 3000v cap. Hopefully its a good one and still has all its smoke (if not, I'll be in touch with Opcom). I'll keep you posted on my progress. If it works with the 807 I may try in on a Knight T-150 that's not getting used.

Dave
KX9DK

[Steve - K4HX]

I would think any overshoot problem would depend on the wiring configuration, inductance of the reactor and that of the modulation transformer.

The overshoot is a function of the circuit impulse response (defined by the values of L and C). The using a large coupling capacitance can result in an impulse response that will create overshoot.

[Steve - K4HX]

I did exactly this with a Viking II. IIRC, I used a 1 or 2 kV oil-filled cap.

Wow!,

Thanks for all the responses and things to think about. Yes, that is the circuit I got the idea from.  http://amwindow.org/tech/htm/obmod.htm

I have an old PA amplifier and a 100 watt audio transformer to be connected backwards that will go through a cap to the top of a 40h reactor. Yes, B+ through the choke to the 807 is only about 600 volts. I did check out epay (thanks for the links WB6IYH). Got an accepted offer on a 4uh, 3000v cap. Hopefully its a good one and still has all its smoke (if not, I'll be in touch with Opcom). I'll keep you posted on my progress. If it works with the 807 I may try in on a Knight T-150 that's not getting used.

Dave
KX9DK

[W4NEQ]

The 3 to 4 safety factor is commonly seen for large mica caps in broadcast RF impedance matching networks - as a conservative approach toward a very reliable system - one exposed to lightning etc., but I've not seen that much margin in HV supplies and audio.   

Also, reading the cited article about the 3 diode negative clipper,  I'm not following the logic there - if the goal is to reduce or eliminate splatter caused by cutting off the carrier, this won't work without some kind of low-pass filtering after clipping.  Hard clipping produces odd harmonic audio products which will pass through the modulation process, widening out the occupied bandwidth.  Hard clipping works well, but only ahead of an appropriate filter.

Good luck with your loudenboomer.  I'm now collecting parts for a 400 watt fet rig.

[k4kyv]

I looked at the coupling cap in my Gates BC1-T broadcast transmitter this morning.  It still has the original stock component, 2 mfd oil-filled, rated at 3000 volts working. The normal full-power plate voltage for the transmitter is 2600 volts, so the Gates/Harris engineers didn't use 3-4 times plate voltage rating; in fact they used less safety margin than I used in my homebrew rig.

[flintstone mop]

I read somewhere that capacitors like to be operated close to their rated HV.

fred

[Opcom]

I read somewhere that capacitors like to be operated close to their rated HV.

fred

Interesting! Because it would tend to press the plates together?

[k4kyv]

I read somewhere that capacitors like to be operated close to their rated HV.

fred

I think that's true for electrolytics.  They tend to "form" to the DC voltage they run at. But I'm not sure if that holds true for oil filled.

Something I have noticed is that oil-filled caps tend to bounce back with a charge, even after they have been completely discharged through a dead short.  That is why it is recommended to store them with a wire shorted across the terminals. I almost ruined my little solid state digital capacitance meter that way.

[IN3IEX]

I would think any overshoot problem would depend on the wiring configuration, inductance of the reactor and that of the modulation transformer.

The overshoot is a function of the circuit impulse response (defined by the values of L and C). The using a large coupling capacitance can result in an impulse response that will create overshoot.

Hi Steve, do we have overshoot even if we apply the same HV to the audio amp and the RF amp simultaneously?
With a high capacitance the heising inductor and the audio transformer can be considered in parallel for audio and for transients. Yes, there will be more DC current in the heising inductor after switch on.
There will be large overshoots if the Audio and RF supply voltages are different, or if the HV of the audio amp is always on, because at switch on the capacitor needs to charge and this will excite a damped oscillation in the L of transformer+inductor.
Anyway I agree that the possiblity of large or small overshoots in this circuit is intrisic, maybe some strategic resistor may help for damping....
 
Maybe under some specific conditions the capacitor could be a "compact" one. I'll try do do some simulations...

[K5UJ]

I did some checking and the Sparta (Bauer) 701, a rig that runs two pairs 4-500A with 3 KV B+ uses a 4 KV 1 uF cap between the bottom of the mod transformer secondary and ground.  What I don't understand off hand is why it is not rated for much higher.  Doesn't the voltage there swing up to over 6 KV with over 100% positive audio peaks? 

[DMOD]

Hi Dave,

I have a 4uF@4000V oil filled cap.

If you really decide you want to use something like this this, PM me.

Phil  AC0OB

[KM1H]

I think that's true for electrolytics.

 

That used to be true when the chemistry used would retain a memory of the voltage first used and would fail if later brought up to rated voltage.

That changed around 40 years ago.

[K5UJ]

This is the kind of thing to look for if you need a blocking cap between the mod. trans. secondary and ground for a typical QRO AM rig:

http://www.ebay.com/itm/New-1uF-4000VDC-High-Voltage-Oil-Capacitor-1mfd-4000V-/130585255542

[k4kyv]

I did some checking and the Sparta (Bauer) 701, a rig that runs two pairs 4-500A with 3 KV B+ uses a 4 KV 1 uF cap between the bottom of the mod transformer secondary and ground.  What I don't understand off hand is why it is not rated for much higher.  Doesn't the voltage there swing up to over 6 KV with over 100% positive audio peaks? 

It depends on whether the cap is wired in such a way as to place the +HV DC across the capacitor. The only case where substantial audio voltage would appear across the cap would be towards the low frequency end of the range, and the capacitive  reactance would be a lot higher than the modulating impedance.  I would say it it is safe to use the same voltage rating for the blocking capacitor as that of the HV filter caps in the power supply furnishing the +HV.

I had never before heard anything about needing it to be rated at 3-4 times the +HV.  I don't think the engineers who designed broadcast transmitters  had either.  My gates runs 2600 volts and the cap is rated at 3 kv.  The Bauer runs 3 kv and the cap is rated at 4 kv.

Does anyone know what RCA and Collins used in the BTA-1 and 20V series?

[W7TFO]

The ITA 1kW AM I scrapped used a 2uF @ 3kV with 4-400's.

73DG