Switching 800VDC with a MOSFET

[KK4YY]

In my continuing effort to make a simple DX-100 complicated, I offer the following circuit for evaluation by the assemblage.

This design keeps the HV "on" during stand-by to reduce the heavy surge of switching the AC line when going to transmit (how important or useful that is, remains to be seen). As the transformer center-tap is effectively lifted by the 'fet during stand-by, there is no return path for the DC that remains connected to the final tubes — like a bird perched on a power line.

K1,R1 is a soft-start. R1 provides a voltage drop on initial power-on and continues during stand-by. This limits surge current as C1 will charge slowly. The value of R1 is selected to maintain the DC voltage at 800V (when loaded only by the bleeder), the same voltage as when under full load of the RFPA during transmit. K1 closes during transmit shorting R1 and the surge is reduced because the filter capacitor is already at full charge of 800V.

R6 limits the current though a zener which provides the turn-on voltage for the 'fet. The 4N25 provides isolation from the HV to the T/R control voltage (12VDC).

Having not built this thing yet, I was wondering if anyone sees a problem with it. Your comments are appreciated. Thanks.


Don

[PA0NVD]

I don't see problems,  simple and clever!! But the HV cap should be well isolated from ground.

[KK4YY]

Nico,

I'll treat the negative lead like a HV lead keeping it away from ground. The HV cap is a non-polarized film unit. I should probably fuse the positive lead with one of those microwave oven HV fuses, just in case the negative lead "finds ground" (shorted 'fet or otherwise).

Thanks,
Don

[w8khk]

36 mA through R6 dissipates almost 30 watts, is that really what you want?

Also I question whether the plate transformer insulation is sufficient to float the center tap at 800 volts?  Of course they are plentiful, so you should not have to pay for a rewind if the smoke leaks out.

[w4bfs]

I agree with Rick .... it is dangerous to lift a ct that high above ground on a non-specified transformer .... a 6146 can withstand high voltage (950 - 1kv) unless it is gassy which makes all bets off anywhoo ..... leave the hv on and bias the tubes into cutoff on standby

[PA0NVD]

36 mA through R6 dissipates almost 30 watts, is that really what you want?

Also I question whether the plate transformer insulation is sufficient to float the center tap at 800 volts?  Of course they are plentiful, so you should not have to pay for a rewind if the smoke leaks out.

I missed R6, that should be 220 k or more, not 22k. Well observed!!. I think that even 470 k should be ok, you don't need more than a few mA through the zener. 
So far I never saw a transformer that had a CT specified or insulated different than the ends of the winding. Many voltage doubler diagrams have also a CT that is high.

[KK4YY]

Rick,

You caught a typo. Not 22K, but 272K (three four 68K/1W in series).


My drawing shows a 4k load. Really, that's the load presented by the tubes during transmit. During stand-by, I had assumed (perhaps wrongly) that there's no load - no current flow through the tubes. With the center-tap lifted, there can't be any current flow, right? But, maybe the tubes look like a resistance, albeit a very high resistance. That would make the center-tap appear to be at -800V with respect to ground, which could be bad news for the transformer and a fatal flaw with this design. Maybe hanging a high value resistor from c.t. to ground would keep the voltage low enough and allow a very small current to flow through the tubes "resistance".

So how did I get here? I couldn't figure out a way to put the 'fet in the positive lead without a lot of hooplah. I was trying to find a simple way. Guess this ain't it, unless I want to see how good the insulation of a DX-100 plate transformer is.

If I put the 'fet in the positive lead I need a higher voltage to turn it on. Maybe hanging a diode at the input of the choke and using a cap input filter might work. That would give me about 1200VDC, 400V above the 800V. That could be a challenge.

Well, back to the drawing board. Thanks for the help with this, looks like I needed it!


Don

[WD5JKO]

There are some commercially made DC high voltage, high side solid state relays from IXYS.

These look like 15 ohms when on at 0.25 Amps load (>= 10ma led current). Look at the graphs, especially surge current versus time.

These are cheap enough to have one for the RF tube(s), and one for the modulator.

$3.67 apiece.

https://www.mouser.com/ProductDetail/IXYS-Integrated-Circuits/CPC1981Y?qs=sGAEpiMZZMvlCQNWg%252Bk%2FqoHt7iADSaDL

Jim
Wd5JKO

[KK4YY]

Jim,

Not bad, but rated at 600V, 1000V abs. max. So close!


But, let me have another whack at this.

I moved the 'fet into the high leg and added a few components but it's mostly the same things moved around.

What say ye all now?


Don

[MikeKE0ZUinkcmo]

Why not just move K1/R1 to the center tap and then you don't need the FET, or, or...

BUT...
I agree with BFS and others regarding probable transformer problems, and suggest you put the tubes into cut-off (ground the screen?).    Use the relay to turn the screens on and off.

[WBear2GCR]

On the first sketchamatic, if the zener opens the little cap goes off like a firecracker...

On the second what voltage are you developing on the high side of R6?

Myself, being a little bit cautious, I think I'd use a voltage divider to generate a voltage
above what the Zener needs, and run a series resistor off the divider for current limiting/
protection...

In general, I'd be cautious about the instantaneous voltage or current drawn -  on the ground
side, current.

[KK4YY]

Bear,

I like your voltage divider idea. I'll incorporate it.

The high side of R6 should be about 1250V to ground. But, I'm looking at it as 450V above the 800V rail to which it's being referenced. But I'm not sure on which side of the choke I should hang its filter cap. I may have it wrong on the dwg. All this just to get 12V to turn on the 'fet!

A key design point is to minimizes current surges. The soft-start circuit should keep surges in check by pre-charging the filter caps through the soft-start resistor. At key-down there's pretty much only the 200mA load of the finals being added. The 'fet is rated at 7A continuous, and for surges well above that. It should only see the 200mA load at key-down.


Don

[WBear2GCR]

The MOSFET wants to see what to turn on?
Is it not sufficient to pull it up to the Drain voltage?

[w7fox]

Don

Very interesting circuit, I like to use mosfets for grid blocked keying.  You may have a problem with the DC soaring with only the bleeder load.  100K isn't going to hold it down.  I have gone to resonant choke input circuits to minimize this and waste less current in the bleeder.  I have had trouble finding  a P channel mosfet that will handle high voltage over 600 volts. 

[KK4YY]

The MOSFET wants to see what to turn on?

The threshold voltage V_GS(th) is about 4 volts. Something like 10 volts will assure an ON state.

Is it not sufficient to pull it up to the Drain voltage?

No. That's the problem. You need a higher voltage. A 9 volt battery could work, it's simple, but not very elegant.

...You may have a problem with the DC soaring with only the bleeder load.  100K isn't going to hold it down....   

The soft-start resistor is to be selected to prevent the voltage from soaring. It will be selected to charge the cap to the same voltage as when under full load.


Don

[KK4YY]

I updated the schematic. I added a "voltage divider" for the zener. It's really more of a voltage limiter (90V) if the zener goes open. If the zener were to fail, it would probably fail short, and the circuit wouldn't turn-on. If the zener fails open, it will probably take-out the opto-isolator. The 'fet has internal input voltage protection (back-to-back zeners).


Don

[PA0NVD]

If Swiched off, the output and the emitter in the opto coupler will be zero, but the collector of the optocoupler will be at 800 VDC, Fatal for the opto
I really do not seen the problem of HV at the CT. The isolation of the CT is the same as the ends of the winding. Your first diagram was ok with 1 - 2 mA zener current

[KK4YY]

Nico,

Dang! This is harder than I'd thought. Putting the 'fet on the high side is said to need an isolated power supply source to turn on the 'fet. I'm beginning to believe it. Maybe a 9 volt battery isn't such a bad idea after all. There are other known solutions, but I have this odd desire to find another way. I should probably lay on a couch and talk to a professional about this (and other) unusual inclinations of mine.



It's my understanding that center tapped transformers do not necessarily have good isolation to the core (which is grounded for safety sake). As such, bridging c.t. transformers, or anything that keeps the c.t. well above ground is risky unless specified by the manufacturer as acceptable (sufficient insulation having been used in the manufacture). For relatively low voltages it's probably okay to do. I had bridged a DX-100 transformer for a project, many years ago, and it survived. This, before I knew it might be a problem. I'd rather not take that chance again, having been lucky the once.


Don

[KK4YY]

Thinking about it, I'm not sure that the DX-100 plate transformer core is grounded at all. The case is grounded when it's mounted. The core is potted inside the case. The core may or may not be attached to the case.

If the core itself is floating, then lifting the center tap may not be as much of a worry. The one that I bridged (circa 1978) worked fine. Maybe it's time to dig/bore/melt through the potting and find out.

[PA0NVD]

The sec halves are normally wound one after the other and all connected to the solder taps all the same way. I NEVER saw a transformer with a center tap constructed so that the isolation is less to the core. Is hard to imagine how you should do that. They are ALL inside the coil former or at distance between insulation layers. Many people hipot a sec to find if the isolation is OK. That INCLUDES the CT. So they are all convinced that there is no insulation difference between CT and ends
You can hipot the winding to check, it safes a lot of difficult circuitry and headaches using your first diagram.
I did switch the power for FET RF generators (+ 70V) using a n-channel FET beacuse there were no P-channel FETS in those days that could handle the current of a 2 kW generator. I used a potted small DC-DC converter block to power the gate switch circuitry, but I doubt if they are suitable to hold-off 1 kV.

[KD6VXI]

Insulate the transformer mounting it on glastic standoffs.

--Shane
KD6VXI

[K8DI]

No, do not insulate the transformer mounting. People who do this increase their risk of shock. If the insulation fails, you WANT it to short out and blow fuses, not invisibly/unnoticeably put a big, relatively accessible piece of metal hundreds or thousands of volts above ground/chassis. Every time I see somebody putting a choke or transformer on wood or plastic, I cringe. Use parts that can handle the voltage. Make sure failures fail safely. Shane, mr no more Millens, of all of us, you should recognize this as unsafe.

Ed

[PA0NVD]

i completely agree with you Ed

[KK4YY]

When the old twist-lock electrolytic capacitors are connected in series, the ones where the case is above ground are mounted on fiber washers and the outer case (negative lead) is insulated with a paper or plastic cover to prevent accidental contact. Same for negative supplies. Similar could be done when mounting a transformer or choke above ground. Too often, this protective measure isn't taken as a cover would need to be fabricated.

I've never had the need to mount a transformer above ground but, if I were to do that, I would make a safety cover for it. An insulating cover or a metal grounded cover should suffice.

In the current instance, I don't intend to create a death trap for myself. I may choose to risk destroying a transformer. I won't risk self-electrocution. Driving to work is enough risk for me.



As Nico alluded, p-channel 'fets aren't available with as broad a selection as n-channel 'fets. This is one reason that I'm looking for a way to use an n-channel device. In the low-side application, that I first proffered, doubts arose due to the use of a center tapped transformer. Had it been a transformer with a single winding and a bridge rectifier, there may not have been any concern about transformer insulation if I inserted the 'fet in the negative lead of the bridge. Perhaps that would have been an application acceptable to all (?).


Don

[PA0NVD]

I do not understand the reason to mount the transformer above ground. You do not trust the isolation to the core but you do trust HV isolation to the primary?. If there is a isolation risk from the center tap but not from the ends of the winding, that means that you can NOT hipot test the isolation of a secondary winding?