The AM Forum

When replacing Diodes on the valiant, I used 1000v 1Amp diodes.  They did not survive even though there is only 600 or so volts on each.  Where's the extra voltage stress coming from?  Inductive kick from the filter choke???

Just curious.

1N5408 RECTIFIER DIODE

3A 1000V

these things are nice, and they are  cheep


klc

I ALWAYS use the 5408s for HV rectifiers. just stack them in series so that the total voltage rating is 2 to 3 times your operating voltage.

I have found that the 1 amp (4000) series diodes will short out if you just look at them wrong. I will only use the 4007s for biass rectifiers, or low voltage control circuits

                                            The Slab Bacon

Hi Ed ...  I don't have a Valiant schizmatic here but as I recall from other discussion that the hv xfmr secondary is 1800V ct (900V either side of center) ... assuming that is so and a full wave ct topology the the piv requirement (again from memory) is 2.83 x Vrms.  this figures to almost 3kV piv ... add to this that you have a choke input supply, this could push this requirement even higher.  the diodes are inexpensive, use plenty of them.  In addition, you might consider a small value snubbing cap across the inductor or to hv return to lower inductor spike generation ... just make sure to check for resonances ... sounds like fun ....beefus

one diode is nowhere near close you should have used 4 per leg. It is voltage not current. 750 VAC X1.41 X safety factor. I used 5 per leg in my TX on the HV and 2 per leg on LV. Single doide on each bias leg.  The rectifier tubes in my V2 had 100 volt drop on voice peaks so a few extra diode drops is nothing.
My HV now sits at 760 and is pretty solid with some extra C in the filter.

Agree with the need for  3 - 4 times the peak.  I'm not good at this but I think it works out something like this: Anticipated RMS E across the diode at reverse polarity X 1.4 X 3 or 4 (safety factor).  So for example 1000 Volts across the rectifier at reverse polarity (when it's not conducting) - the peak E is 1400 volts -- apply 3 or 4 times that value for safety factor and you get 1000 x 1400 x 4 = 5600 PIV rating = plenty of safety factor.  3 times would get you 4200 volts - that might be enough - but diodes are cheap -- go for a 4X safety factor.

73, Al - Violet Toilet Paper - W1VTP that is

One of the other posters said:

"I don't have a Valiant schizmatic here but as I recall from other discussion that the hv xfmr secondary is 1800V ct (900V either side of center)"

(I know there is a nicer way to quote something, but I don't know how to do it.)

Anyway, I plugged these parameters into a power supply simulation program and then put a simulated resistive load on it of 2K Ohms. Not knowing the size of the choke or filter capacitor, I assumed 10H and 40uF. I assumed a choke input single stage filter.

The results were:

Peak reverse voltage across the diodes:  2544v

Peak diode forward current during the startup transient:  1.56A

Peak repetative diode forward current during steady state operation after startup: .463A

Output steady state DC voltage from the supply: 786V

Steady state DC output current from the supply: .393

There is also one other interesting parameter that may partially explain the short life of filter capacitors. The peak voltage across the filter cap during the startup transient was 1220V. (This was also the peak output supply voltage in the choke input filter).

While this peak only lasted a cycle or two, still, you need a filter capacitor with plenty of reserve voltage capability.

BTW, a soft start reduces the startup peak current on the diodes. It also greatly reduces the startup voltage spike on the choke and on the filter capacitor. Plus it also avoids subjecting the whole transmitter to the 1220v spike.

I agree with the plenty of safety margin. So given you need diodes that can withstand 2544V reverse voltage in this simulation, just to be safe you should use at least 5KV or better yet 6KV on each leg.


Dave

The QUOTE button on the upper right of a post will quote the whole of the text, and bring up a post window. Delete any text you don't need between the "QUOTE" statements. Like this:

Did you include the series resistance of the choke? This will tend to damp down the peak some. Also the series resistance of the transformer will damp the peak surge too.

Very nice detail...

There is a reason why Johnson didn't use 816 rectifier tubes which only have max inverse rating of 7500V. They used 866A tubes which have a max inverse rating of 10,000V.

I think using even 6KV diodes is going a bit light. I'd go with the full 10KV ratings.

Also, what happens in your simulator program if you throw a big .1uF cap to ground right at the output of each diode? I always did that to help minimize the problem.
[/quote]

OK, I tried that. (Actually I changed the circuit to a capacitor input filter with a .2uF input capacitor which amounts to the same thing at least as far as this program is concerned.)

This didn't seem to have much effect:

Peak voltage spike across output (40uF) capacitor = 1217 V

Peak startup diode current = 1.68

Peak reverse voltage = 2566 v

So you get a little higher peak current as one would expect as you have to charge the .1uf capacitors starting from zero during the initial cycles.

What really, really helps is a soft start.

I wasn't that convinced that soft start mattered that much when you use diodes (which can stand very high non repetative current pulses), but after fiddling with this program I plan to use some kind of soft start from now on.

Does anyone have experience with negative temp coefficient termistors for this application, BTW.

I also was poking around in an old BC-610 maintenance document recently and they mentioned that during the later WW2 production models they put a spark gap across the HV power supply choke to short out power-on transients. The were having problems blowing chokes -- and these were mil spec 11H jobs with 10,000 volt ratings. The circuit used a FW center tapped design with 866as or 3b28s depending on production date -- and no soft start. Plus incredibly huge power transformers which were way over rated and probably had very low DC secondary resistance -- compounding the startup surge problem.

BTW, in this simulation (without the .1uf capacitors) the terminal-to-terminal peak voltage on the choke is 1200 v . Another reason for soft start.

One of these times I get the quote thing right :-)

I used 31 Ohms for the transformer and 30 Ohms for the choke. Probably too low for the choke. Those are parameters in this simulation.

One thing I'd like to try, but can't because the program won't support it, is to put capactors across the diodes -- perhaps like .01uf. I've seen that done and wonder if it might limit current spikes. Probably wouldn't have much impact on PIV as that is mostly determined by the waveform of the input (sine wave in this case).

Another observation is that using a capacitor input filter with the input capacitor being somewhat similar to the output capacitor (ie a PI circuit) is greatly easier on chokes as both ends of the choke are "tied down" by capacitors so the voltage spikes across the choke are limited.

Another point is, that I was considering using a 10H 1400 v choke in a 3KV supply and putting it in the negative lead figuring this would limit the voltage stress to within the ratings.

Bad idea. The simulation shows terminal-to-terminal voltage on the choke of way over 1400V so even with one end tied to groud the choke would be stressed beyond it's ratings.

Well, simulations are no substitute for good engineering practice. All simulations are only as good as the underlying mathematical models and they are only that good if the software developer doesn't introduce bugs.

And the best models only partially account for reality.

This program isn't even close to sophisticated enough to deal with hash and oscillations from MV tubes for example.

Another factor no one ever talks about is the impact of the inductance of the plate transformer. Although the voltage is determined by the turns ratio, and the resistance of the winding is accounted for by the model, different transformers could have different inductance on the secondary winding depending on the core specs and the total number of turns. It seems logical to me secondary inductance might have a significant impact on transients also. But this program has no way to take that into account.

I bought 60 5408's for the Gonsets - those damn 1n4007's even look scroteless. with the 572B's I don't see how they would survive a black james bond rap transmission. I believe I'd see smoke and then i'd be off the key. ifn I wanted to be or not.  ::)

The 5408's are well matched in reverse impedance that I am told that you no longer have to use balancing resistors on the series stacks. Dunno- I still put 470K half watts across them. By the way, I did up the Apache with them and used two on each leg on the MV and they blew after a couple of weeks - I put 3 in and was good to go. Use the advice in the thread! Mo is betta.

personally, I'd rather have 3B28's.

I'm not reading all responses, so this might be a dupe of what others' have said...
But.. When calculating the PIV rating for the diodes, remember that the charged capacitor voltage is
additive to the peak reverse voltage on the secondary winding.  I'm assuming this a conventional
full wave circuit?

Pete

Well,  I have met Hammy Hambone and he is ME...

I have 10 1n5408's and a couple more HV electrolytics on order from mouser. 
I figure that the 5kV per leg should be plenty.

If not, I guess I'll put 3B28's in there and call it DONE.

ed,

3B28's are he schiznit. dont be doing what I do an think everything has to be just so. It's a ugly trap.  :-\ you never finish anything.

Ha, I've been dogding that "trap" for years now.... 

Anyway, I do kinda feel silly on this one.  I had 3B28's available, but I pulled the 866 and sockets out of the rig to SS the power supply.  Prolly would be Fat Dumb and happy with the 3B28's and running great.

I think I will use the tubes for the HB rig I am working on.

2400v on the plate. Should be niiiice...

The RCA tube manual indicates that the 3B28 has a peak inverse voltage of 10K and says that it can be used with a full wave center tapped transformer of up to 7000KV to produce 3500 VDC before the filter -- assuming a choke input filter.

So tubes don't need nearly as large a safety margin as do silicon diodes. They can withstand very short duration voltage spikes better -- or so it seems anyway.

and they used to be 2 bucks each at hamfests, NIB. dunno what they are now.

I am installing 1N5408's into maul #1, knowing that when Slab speaks, it's best to listen.

Always a Wise thing to Listen to Slab,

Not trying to High Jack anything here but me thinks a section in the AMfone's Online Handbook could use a nice treatise on this subject matter. This very idea on How To and with What...

I Looked in a coupla headings over there and found nothing...anyone agree.?