Plate transformer secrets revealed.....

[N3DRB The Derb]

I hate transformers where you have to take it off the chassis to find out what the dope is.  I should be glad it's marked at all I guess.

I'm assuming that voltage marked there is the total since the max you can give a 4X250 is 2KV.

No current rating marked but it used a pair of 816's for rectumfiers = 250 ma.

has anyone seen a unit like this or know who made it? Maybe Wile E. Coyote ordered it from ACME? nah, Acme always marked their stuff. 

[k4kyv]

Is the entire winding 1750v, or is that the voltage each side of CT?

[N3DRB The Derb]

I think I'm going to have to measure it to make sure. It' s the plate xmfr out of the 10 buck MARS transceiver I got recently. Used a 4X250 final, 2 816's, so I'd bet thats the voltage.  After everything else and adding a bleeder, 1600 volts I'd guess. Couldn't ask for better for 811A's. Little light for 572's, but the tubes would last forever.

[k4kyv]

If that is the total  winding voltage, you could take a chance with the insulation rating at the midtap and run it with a full-wave bridge rectifier.

[N3DRB The Derb]

well, crap. I removed the filter cap and it's only rated at 1KV. So I have a low voltage hi current job, unless they just Js'ed anything they could find in there. I'm gonna have to measure it.

I need to throw a solid 2 KV at these gammatrons to get good pow-pow-power out of em.

[WD5JKO]

  So it looks like you have 1750 CT transformer. Since the CT has a large ceramic insulator, I'd bet you can float it like Don suggested, and use a full wave bridge rectifier. Then go capacitor input filter, and you get about ~ 2500v no load, and it will likely droop to about 2KV loaded. For Cap input, use a 2 step turn on with a relay to diminish the turn on surge (only good if HV stays on), or use a zero crossing solid state relay (20 amp or so) if you plan to "push to talk" (PTT) the HV.

   You could have a HV switch to use the CT as a "Tune" position where the B+ will be 1250 unloaded, or 1 Kv loaded.

 Go for it........

Jim
WD5JKO

[KM1H]

The first thing Id do before trying a FWB is to measure the full secondary resistance then go calculate the instantaneous charging current and voltage peaks. This is fully explained in older handbooks. With just a fat C its very easy to have a runaway supply and smoke. Been there done that many years ago. In some cases the only way out is a input choke with an acceptable minimum L.

The 816's should have told you its not a HV xfmr.

Carl
KM1H

[N3DRB The Derb]

pair of 816 can do 2400 volts to the filter at 250 ma. I had hopes of a wee bit more voltage, but it will make a dandy plate xfmr for 811A's which I have 4 of here.  So I guess those will be my mod tubes. There was also a non-descript choke there with the WW2 USN anchor symbol on it. 108 ohms dc resistance.

[Opcom]

please consider this for those pesky center tapped transformers:
http://amfone.net/Amforum/index.php?topic=17807.msg122938;topicseen#msg122938

[WD5JKO]

> With just a fat C its very easy to have a runaway supply and smoke.

Reply by Jim, WD5JKO;

   Please define runaway for me. I think I know what you mean though..

Sure the initial turn on surge if instantaneous at peak of the voltage input waveform will be sky high, limited by winding resistance (pri and sec), and transformer leakage inductance. To avoid this we use a step start relay where we initially add something like a 10 ohm resistor in series with the primary for a second or two until the relay closes.

The problem with a step start is it doesn't work well with push to talk unless you key and talk for an hour, Ozona Bob style.

An alternative I didn't mention was to switch the B+ and leave the power supply on. The switch however needs to be pretty special to switch 2-3 Kv like a vacuum relay.

I did mention a solid state relay (SSR) if the user wants to PTT the power supply. This will guarantee that the AC primary is turned on at voltage zero crossing. This way we go from zero volts to full b+ in several line cycles when the current always starts at zero. No more hitting it with just a mechanical relay where you hear that wicked clunk and transformer groan; and then the next time you hit it right and you only hear a clunk, but no groan.

The problem with SSR's is that whether they use a Triac or back to back SCR's, they work best with a resistive load where there is continuous current drawn over each line half cycle. With a power supply using capacitor input filtering the current drawn is spiky and brief starting just before midpoint of the half cycle. In order to make the SSR behave, we must provide a minimum load over the line cycle. Many BA rigs use a 120v incandescent lamp indicator for HV present. The Viking 2 and Globe Kings do this. Having a lamp such as those across the transformer primary should make the SSR behave.

The bad news with capacitor input filtering is that the current magnitude is high leading to high I^2 * R losses, and the good news is that those losses are brief and when integrated over the line cycle, are not that bad.

On the subject of thermal runaway, I see how this might happen. Lets say the transformer was 1000 vct at 1 ampere. With FW CT rectification, choke input filter we get about 450 VDC at 1 ampere, or 450 watts of B+. Take that same transformer, put a FWB on it with capacitor input, and we get roughly (depending on winding resistance) 1200 vdc at about 400ma. If we dared to increase the load to 1 ampere, we would indeed have thermal runaway, and eventual smoke.

When I suggested FWB, I did not mean with 816's since these are rare, fragile, you'd need 4, and the worst is that you'd need 3 filament transformers. Instead I'd stack several 1N5408 3 amp, 1000 V PIV solid state rectifiers on each leg of the FWB. Since these are < 50 cents each from Mouser, maybe six on each leg, or 24 total. You could use less, but they are so cheap the extra few won't break the bank.

Jim
WD5JKO

[KE6DF]

On the ppt/CW keying issue, here is another somewhat left field idea.

See the attached jpg file.

It's a page from an old Thordarson catalog. It shows using the 22R50 phase shift transformer in a circuit with Thyratron tubes to provide a variable high voltage plate supply.

The idea is that by varying the phase of the signal on the grids of the thyratrons you can pass through part of each cycle rather than whole 1/2 sign waves. (The thyratrons act as switched rectifiers.)

Once the resulting rectified pulse stream is filtered you have a variable voltage output depending on the phase shift.

So you end up with a variable HV supply with pretty good voltage regulation -- as good as a choke input supply anyway.

As a side effect, they also include keying the plate supply by keying the grids of the thyratrons.

They claim a very nice CW wave form using this method. Would work for PTT also

I happen to have a 22r50 and a few 5557 thyratrons (roughly 866's with a grid) so I may build such a supply at some point.

[Opcom]

That's a very decent approach.

[WA1GFZ]

When converting a FW CT to FWB the current rating drops in half. That's if the CT doesn't flash over. BTW the CT is now at 1/2 DC output voltage.
But VA is VA and there is no magic way to increase the power rating of a power transformer. You can pull a little extra power if you can live with the voltage drop under ICAS conditions

[N3DRB The Derb]

I am a swinging choke man myself. I'll always build a 2 choke PS even for a linyar.

[KM1H]

Generally the 816 is not used at much over 1200V as they have a reliability issue if run at the full spec.

With only 125ma per tube MAX its easy to blow it apart with the instantaneous charging current of a C only filter. With no load the HV soars and zap goes another MV.
Step starts were used as far back as the mid 30's but it doesnt always work with a MV tube already at the edge.

A choke input still has to meet the minimum L requirement otherwise the tube still arcs when the PS is unloaded.

A bleeder helps but with just a C its going to be eating up a big part of the available current and its still a power hog even with a choke when using a high impedance xfmr.

Carl
KM1H

[WD5JKO]

 

I am a swinging choke man myself. I'll always build a 2 choke PS even for a linyar.

  I'd reconsider unless you have a large capacitor as part of the last filter section. Otherwise the power supply filter resonance will ring when the load is applied, and released. So using a L-C-L-C filter like in the old handbooks where the chokes might be 10 Henries or more, and the capacitor might be 4 mfd at most will cause transient load change ringing where the amplitude peak can approach the DC level.

  I have seen many posts on AM FONE where folks beefed up the output cap on many a vintage rig. Every post says good things about doing so.

  If I made a FW bridge out of four 816's, and three filament transformers, I'd have a neat conversation piece for sure. Sorry, but MV tube flash over, expensive MV tube prices, and spotty availability makes the choice a no-brainer for me to stack several 1N5408 diodes that can can be had for as little as 20 cents each.

   For those that like the 816, remember that the 125ma rating is the continuous rating. In a FW-CT power supply with dual 816's, we can pull 250 ma from the power supply output because each tube pulls 250 ma for 1/2 the time, or 125ma average. If we use four of these in a FW bridge, remember that the PIV rating is doubled. The 15 volt 'ON' conducting voltage will also double since two tubes forward conduct at a time.

   The modern power supply with step-start, beefy SS diodes, low loss transformer, and a big CAP (say 50 MFD at 4 KV for a linyar) is my choice so long as I have a nice bleeder on it. In fact if this power supply is for a big AM rig, the RF final will steady the load, and the modulator will have a very stable voltage to work from.

    I might really upset folks here if I said using the same parts I mentioned above, but instead use a full wave voltage doubler configuration. Virtually all modern linear SSB amplifiers do this; even the really good ones.

73,
Jim
WD5JKO

   

[N3DRB The Derb]

I'm not going to use my 816's. Too light duty for me. Never said I was going to use them. thats just what was there when I got the rig. I like 3B28's and thats what I'll use in any PS I make, yo. 

I'm sending them out to KC so they can reside in here: lookit the picture and you can see they're going into the house of Millen. Aa nice place to live. 

[KM1H]

For those that like the 816, remember that the 125ma rating is the continuous rating.

Thats only at 1750V DC. Above that its steadily derated.

1N5408's are good, 6A10's are mo betta.

Carl
KM1H

[WD5JKO]

> 1N5408's are good, 6A10's are mo betta.

Maybe so. I looked at the specifications, and the reverse current ratings at rated voltage are pretty different, as is the junction capacitance. The 1n5408 is rated at 500 ua reverse current at 1KV and Tj = 150C. The 6A10 is rated at 100 ua reverse current at 1KV and Tj = 100C. The old 1N4007's circa 1980 would die at 1ua once avalanched. I hope modern 1N4007's are better these days.

I haven't seen reference to "controlled avalanche" in years within diode datasheets. That said, this is important. Also I don't see any reference to reverse recovery time (Trr) either. The typical general purpose silicon HV diode might take 10-500 us to turn off. The consequence of this is in a Full Wave CT arrangement the transformer is shorted out during the diode reverse recovery time, and when the slow diode finally shuts off, the transformer leakage inductance recoils a transient that might break down the diode(s). The fix is an R-C snubber, or faster diodes. The slow diode issue in a hi fi amp can hurt dynamic range from the back emf periodic spikes getting into low level stages, and since it might be only 100ns wide, it is very hard to filter out.

  The audiophiles boast about Hexfred diodes........These have a Trr of about 50 ns.

If you look at a HV solid state supply with GP HV silicon diodes with a scope, the Trr issue is very apparent, and the spike into a choke input filter comes right through unless it is suppressed up front with a suitable R-C snubber. I once got rid of the spike by resonating the choke where I had a choke input filter.

Whether the 6A10 is better than the 1N5408 in HV boatanchor equipment remains unclear to me. The data sheets leave out all the good stuff. 

The 6A10 does appear to be a mule though. That said, if it punches through and shorts out when reversed biased beyond 1 KV, then it might not be the best choice.

Jim
WD5JKO

[Opcom]

just pile up some silicon carbide rectifiers. no recovery issues.

Derb man I have to say you have some of the nicest stuff!

[KM1H]

I agree the 6A10 could react as you suggest in a FW CT but that doesnt happen in a FW doubler. Snubbers have been suggested for decades in a CT design.

The main advantage of the 6A10 is the surge ratings which makes it darn near bullet proof no matter where on the cycle the PS is turned on without a step start. I suggest that more diodes fail from power line surges and lightning than what happens on the DC side.

The guys running serious power and a couple of hundred uF of C swear by them for complete reliability. A step start is used for tubes with handles.

Not a bad deal at around $.30 in small quantities.

Carl
KM1H

[KE6DF]

There type of prepackaged rectifier stacks (at least that is what I assume they are) are another option:

http://cgi.ebay.com/30kV-2A-High-Voltage-Diode-HV-HF-Rectifier-Tesla-Ham_W0QQitemZ150299747551QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item22fe9024df&_trksid=p3286.c0.m14&_trkparms=65%3A12|66%3A2|39%3A1|72%3A1205|240%3A1318|301%3A0|293%3A1|294%3A50

Most are from China.

Has anyone had any experience with them or similar units?

Personally, I like playing with tubes. I have to admit I've never built a 1KVA plus power supply, but I have some 4B32's i'd use if I did.

Seems to me the xenon filled tubes (3b28, 4b32s) were the top of the rectifier tube development curve.

[Opcom]

best to know more about the manufacturer of the diodes:

http://www.hv-semi.com/index.html

http://www.hvpower.cn/