HB Transmitter RF Deck

[W7SOE]

Finally got around to sketching a rough schematic of the RF deck. (PDF Attached)  I would appreciate any help in understanding what is going on .

To my inexperienced eyes, I see:

Two 813s, parallel, grid-driven.

What appears to be a clamper circuit (the 6y6 and 0d3).  This circuit is confusing with both the grid and screen currents involved.

The band switch is a turret style coil switch (B&W), combined with the variable cap at "E" I think this is the input tuning.  (Shown here as a box on the lower left)

The caps at the "B" and "C" inputs are ganged.  One, I believe is "grid tuning" (whatever that is) and the other goes to two porcelain insulators located at the bases of the 813s.  Perhaps these are for neutralization plates that were not needed?

The Plate and Tank circuits look straight forward.  The antenna loading and capacity adjustments are new to me....

I hope the schematic is mostly correct if not complete on the input side.

Thanks

Rich

[W3RSW]

Ok Rich, in reference to your email, here's the explanation.
If you can obtain an old "Radio Handbook" or any "ARRL handbook" from the 60's or earlier you'll see you have a standard circuit of the era.

What you have is a traditional dual 813 rf deck with pi net output, extra and switchable loading capacitors in the output of the pi and if you've drawn that switch correctly, it progressively loads up the additional loading capacitors in parallel for at least two positions.  The first coil off the plate is probably for 10 meters, the rest of the taps and add'l loading caps. good for bands to 80 meters.

Your screens are 6YG clamp tube protected, that is grid bias is normally established by an input signal of about 15 to 20 watts which causes current to flow through a grid dropping resistor, hence making the grid about minus 160 or -170 volts with respect to the cathode (filaments) of the 813's.  If there is no grid drive then the clamp tube keeps the 813's more or less cut off.  Without the clamp tube, during loss of drive in AM or in CW service when the drive is cut off, the 813's would have no negative bias on the grids and hence conduct way too much plate current, "run away and melt down."  The OD3 cuts the screen voltage to zero when grid excitation is removed, not just a somewhat but still a safe current if it's not in the circuit.  You want to be sure the OD3 goes dark, no current, when grid excitation is removed.

I need to see a little more of your turret circuit to make other than a conjecture of your grid input circuit.  What are the approximate values of the capacitors shown? This will help identify the lash-up.  Usually a neut. cap. is very small, few plates or only two, wide spacing.  The grid tuning will have many plates and is usually around 100 pf.  The turret probably has link coupling, a few turns on the 'ground' side of each coil from the input jack and many turns on the grid side.  The grid side of the, say 100 pf, cap in parallel with each coil will be above ground. There will be around 300 to 800 pf from that lifted ground to real ground.  The neut. cap. will feed from the plate back to this lifted ground in order to make the neut. "bridge" complete.  You feed rf into the amp. without any HV and tune for minimum output as you adjust the neut. cap.  It may already be set up and close so don't fool with it at first. You'll need to get some instruction and practice here. 

Also, you'll get better linearity if you feed the screen dropping resistor from the HV on the non-modulated side of the mod. transformer.  Your schematic shows the screen dropping resistor coming from the modulated side, same line as feeding the plate choke. You'll need another HV jack and wire run to the HV power supply to do this. What you already have will work but has a little more distortion than the two lead HV I've described.

Anyway you have a classic; it'll shine when you get it on the air.

 

[WB2RJR]

Rich,

That Clamp tube circuit is a dead ringer for the one shown on pg. 154 of my 1963 ARRL handbook under protecting screen grid tubes.

The amp appears to have grid leak bias (no seperate grid bias supply) so the bias is developed  by grid current flow when the amp is being driven. If you lose drive the grids will no longer have negative bias and the 813s will conduct like a mother.

So bias from the PA grids is applied to the 6y6 grid and this biases the tube off, no conduction. If drive is lost, (as in keying an earlier stage) the grid on the 6y6 goes to zero it conducts dropping the screen voltage and turning the 813s off. The VR tube is used to completely remove any screen voltage from the 813s.

I hope that makes sense. Remember I'm a petroleum geologist not an EE.

Marty WB2RJR

[W7SOE]

Thanks Rick and Marty,
    I know the risks of exposing my inexperience here but I have some basic questions:

Where does the negative grid bias come from?  It looks to be provided by the RF exciter, but is this not a sine wave at RF frequencies?  How does this become a negative DC offset?

The clamper:
I see this as the input RF biases the 6Y6 grid to a negative voltage cutting off the tube.  In this state the 0D3 regulates a positive voltage to the screens of the 813s.  When RF is lost the 6Y6 is turned on, the 6Y6 plate voltage drops, and the 0D3 is off.  Is this even close to being correct?  Where would the 813 negative screen voltage come from?  I assume it would have to be negative to cut off the tube.

Rich
 

[The Slab Bacon]

Rich,
       the way that the clamp tube is supposed to work is: When rf drive is applied to the final tubes some of that rf is rectified in the grid circuit creating the negative operating bias accross the grid leak resistor. This cuts off the clamp tube and the amplifier acts as normal. When the incoming rf drive falls off the bias goes away  the clamp tube starts to conduct. The conduction of the clamp tube "clamps" down (shunts) the screen voltage to the final amp to a level that limits the run away plate current of the final tubes to a level that wont cause them to melt down. Without any negative biass at all it is sometimes difficult to get a tetrode to cut off by just clamping down the screen, but it will keep it at a safe level if adjusted properly.

Grid leak biass was the way to go for many old timers. I prefer to have some fixed protective biass as well and eliminate the clamp tube. But that is my preference and it requires an additional power supply for the bias.

Johnson Rangers have only grid leak bias on the final, and a clamp tube. You actually have to drive the final into class C operation. You can actually watch the output come up and the plate current drop back as you increase the grid drive to the final. Its kinda neat to watch.

                                              The Slab Bacon

[AB2EZ]

Rich

1. Where does the negative grid bias come from?

Your attached .pdf doesn't show the details, but the r.f. input signal is capacitively coupled (the coupling circuit doesn't pass DC) to the grids of the 813's. The grids of the 813's behave like diodes, which conduct in the forward direction when the grids are positive with respect to the cathodes of the 813's. Therefore, on each r.f. cycle of the input signal, some grid current will flow in the positive-going portion of the cycle (near the peak), but no grid current will flow in the negative-going portion of the cycle.This imbalance will add some net charge to to input coupling capacitor on each rf cycle. As the capacitor charges up, a voltage accumulates across it (V=Q/C), until that voltage is large enough that the current flowing through the variable resistor (attached to the grid meter) is just enough to balance it out. I.e.,  the resistor current, V/R (which discharges the capacitor throughout each rf cycle) balances the grid current of the 813's (which charges the capacitor during positive peaks in the input rf signal). If it helps... you can think of the combination of the capacitor and the diodes looking into the grids of the 813's as a "peak detector". Again, the resistor removes charge from the coupling capacitor throughout the duration of each rf cycle, and the grid current flows only during the positive peak of each r.f. cycle. Therefore, if you were to look at the voltage across the input coupling capacitor, it would take a small step up during the positive peak of each rf cycle (corresponding to the flow of grid current causing the capacitor to charge up), and then it would decay back, to where it started, for the remainder of the cycle (corresponding to the flow of current through the resistor). It would look just like the voltage across the filter capacitor of a half-wave rectified, capacitor-input power supply... which takes a small percentage step up at the peak of each AC cycle (when the diode conducts), and then decays back to its original level (due to the load on the power supply) during the remainder of each cycle. 

2. The screen circuit

If the grid drive (i.e., rf input signal) goes away, then the 6Y6 will turn on, dropping the voltage on the plate of the OD3 (with respect to the cathode of the OD3) below the "sustaining" voltage of the OD3. I.e., the OD3 will stop conducting.  When the OD3 stops conducting, the screens of the 813's will be pulled down to ground potential by the resistor at the bottom of the OD3. [Note: that resistor is connected between the screens of the 813's and ground). With the screens of the 813's at ground potential, the plate current will be low enough so as not to damage the 813's.

The OD3 does not "regulate" the voltage on the screens. It simply adds 150 volts of voltage drop (when it is conducting) into the series path between the modulated plate voltage and the screens of the 813's... which helps to ensure that the screen voltage to be less than the plate voltage (as is desired), even if the screen current is zero.

Observation: In my opinion, there are better circuits to use... particularly if you plan to plate-modulate the 813's. The fixed voltage drop of the 0D3 is, in my opinion, going to produce a limitation on the achievable linearity of the modulation. In fact, in my opinion, the use of the OD3 is an unnecessary, "brute force" way to protect the 813's from loss of grid drive.

Best regards
Stu

[W7SOE]

Ahh, I think it was the rectification in the grid circuit that I was not seeing.  Still hard for to me to visualize.  How rectified is that signal?  What does it look like?  The original sine wave but downward shifted?

I think my JVII has a similar arrangement to your Ranger.  I will try to observe the clamper "releasing" the finals as you describe.  I suppose I would shut off the oscillator and increase the grid drive while watching the plate.

I have adjusted the clamper on it so I know it works.   

Thanks

Rich

[W7SOE]

Thank you Stu, excellent explanation.

Being the dork that I am I simulated an ac source charging a cap to a negative voltage using spice.

Looking at my schematic I am not sure which cap(s) are the input caps, must be those to .001uF caps......

Does this negative bias carry through to the grids themselves?  If so does this set the operating point of the tube?

Rich

[AB2EZ]

Rich

The input coupling cap is not shown in your schematic. It could be in the box labeled "Band Switch" , and, if so, will be in series with the r.f. input signal, and somewhere between the r.f. input signal arriving at port D and ports A and B, ... or, it could be assumed that the incoming rf signal is already AC-coupled. In any event, all that is necessary is a series coupling capacitor (therefore DC-blocking) somewhere between the RF input and the grids of the 813's. 

The coil going from port B to the 6Y6 grid is an RF choke, which passes the average current needed to discharge this capacitor. The two .001 capacitors (strangely shown in parallel in your drawing) are r.f. bypass capacitors, which work in conjunction with the RF choke to keep RF off of the 6Y6 grid.

The negative bias is, of course, on the grids of the 813's... biasing them negative with respect to the cathodes of the 813's. According to the RCA specification sheet for the 813, for plate modulated, class C-operation the DC grid voltage should be around -175 volts. This implies that the rf input drive signal (as it appears on the grids of the 813's*) will have to be over 175 volts in peak amplitude (actually shown as 300 volts peak amplitude on the data sheet). On positive peaks, it will drive the grids positive with respect to the cathodes... causing grid current to flow... therefore placing some net charge on the coupling capacitor.

*If you were driving this amplifier with a 50 ohm source, providing 20 watts of power, then the peak voltage of this input rf signal would be ~45 volts. The input circuit (inside the box labeled "Band Switch") would have to "step this up" to a higher voltage, and a correspondingly higher impedance level... e.g., 300 volts peak at an impedance level of 2200 ohms. This is like a typical tank circuit, in reverse: low impedance and low voltage in vs. high impedance and high voltage out.

Stu

P.S. As a friendly suggestion.... building an 813 amplifier may not be a good place to start... because the voltages are high enough to kill you if you make just one mistake... and the power levels are high enough to destroy a lot of expensive components in the blink of an eye.

[W7SOE]

Stu,
   Thank you again for the help.  This rig is already built, I am just trying to understand it before I fire it up.  (In order to avoid the expensive flashes you referred to)

I am well aware of the HV, scares the crap out of me.....   Every time I open the back I touch a grounded rod to about everything I see :-)

Rich

[W3RSW]

Great explanations; these guys understand the rigs they build or operate.  They sure know their stchufff. What a service!  Makes one proud to be a ham.

Couple of comments. 813's like about 13 to 15 mils. grid drive.  So with that current for two tubes from the self rectification action described, say 30 mil. through a 5k grid resistor, voila, you have -150 volts; through 6k, -180 volts.  So the variable resistor shown easily allows setting grid voltage at operating conditions to -170 or so.  As mentioned, be very careful here. Turn everything off, discharge the HV caps, etc. before making adjustments.

So here is the time to mention making a discharge 'stick' so you can safely discharge HV filter caps. without damage.
The experts can help here, but basically it's a 3/4" to 1" dia. highly insulated material "stick" (good broom handle or fibreglass) about two or three feet long for safely poking the HV terminal to be discharged with a 10 to 50 ohm, 20watt resistor wired from a point tip of the stick to an allegator clip that can be clipped to chassis or proper ground depending on circuit or component to be discharged.  Everybody has their favorite combo so I'll defer to others at this point, but you get the idea.  Safety First.

Also the guys described a combination of fixed bias and self bias.  Technically the combination of the two might also be adjustable for least distortion.  But half and half is pretty much as good as Joe Ham can hope to get without too much sweat.  So you're looking at a regulated bias of -75 from a 40mil. voltage reg. tube with the additional -75 to -95 being made up from self bias from input driving power through a grid resistor of say, half the value required originally if using full self bias as is built in your rig.  The half and half combo does require a separate grid supply though but it will appreciably cut down some distortion.  The first -75 also acts as sufficient to protect the 813's without the clamper tube circuit.  

Also you show pin 5 of the 813's both grounded directly and with a 'parallel' cap. to ground. Confused here since direct connection to ground should short things very sufficiently.  Simple typo in the drawing?

Just one more thing, you might have a varible neut. cap. in the bottom of the lifted, above ground grid circuit I mentioned in first post.  It will then have many plates vs. a more normally seen small, few plate neut. cap. ordinarily seen going from plate to 'lifted' RF ground.  That might account for the three cap's seen in next to your "black box" turret circuit.  Then the other part of the neut. bridge is simply the two feed thru's seen sticking up through to the top of the chassis next to each 813.  The peekaboo screw/nuts might have been enough at a few pf, but I think if that's the circuit, those peekaboo's originally had a small vertical metal plate attached, say 1/2" x 2" long about 1" away from the 813's.

Also you might have a real input loading capacitor of high value, say 300 to 500 pf, in series with the few turns of the input link coils.  I'd really like to see a good pix of the bottom of your RF deck sufficient quality and angle/ angles to see the lash up.  At any rate I seem to see one capacitor too many in the input circuit. Vellly intellesting...

[W3SLK]

Rick said:

So here is the time to mention making a discharge 'stick' so you can safely discharge HV filter caps. without damage.

Personally, I don't like the "broom stick" especially if it has varnish or it has been saturated in oil. I don't fart around. I got a Government type Shorting rod that I used in the military. You can usually find them at hamfest for a song. They have a derating curve on distance vs. volatge. I think they are actually rated up to 35KV. Plus they have a gorilla like clamp at the end. Unkle Sam is pretty good about his safety schticks.

[W7SOE]

Thanks again to all for the help.  Ready to crash but I wanted to get a picture off, the least I can do after all the help here.  The picture is labeled:

A-  Antenna Loading Var. Cap.
B-  Caps and RF choke at Modulated HV input
C-  Capacitors associated with "Capacity Switch", at output
D-  14k 25W resistor between the HV input and the 6Y6 plate
E-  RF input tuning cap
F-  Grid tuning cap
G-  Neutralization cap* (ganged with grid tuning cap)
H- grid bleed resistor
 
* This is connected to the feed-throughs....  Perhaps these used to have plates attached.  You can see these feed-throughs in the picture of the top of the RF deck.

Perhaps this picture is helpful in light of sloppy schematic.

Rich

[W3RSW]

Nice pix but I still can't see the wiring to the turret.  The original builder did fine work. I sure like his color coding of filament green, screen yellow, etc. wiring.

You might have a very traditional, but little used anymore, split stator neuturalizing circuit on the grid. 
An example is on p. 163 of the 1962 ARRL handbook.
You really need to open up the turret shield and trace this out.  For my peace of mind if nothing else.

Add'l note to previous self rectification discussions:
In self rectification and creation of negative bias, grid with respect to cathode, an 'input' or separate 'storage' capacitor as such is not required.  The tuning cap. in the cap./coil parallel input tuned circuit provides the flywheel.  This is seen in just about every diode receiver detector ever built, e.g., direct connection from top of parallel resonant circuit to grid.  Bottom of resonant circuit without neuturalization can be simply connecting to DC ground through the grid resistor of 5 to 10k previously mentioned, and to RF ground through a capacitor.  Sometimes a neut. circuit is ommited at lower powers especially if a loading resistor of 5k or so is placed across the resonant circuit to tame it down.  I see in the pix a VHF choke is in the grid circuit as well as the more usually found plate circuits. Wonder if the orig. builder ever needed much neuturalization, found it stable enough and didn't bother to rig up some aluminum plates topside?  Or perhaps the peek-a-boos of the screws looking up provided a few 'pf' and was sufficient.

[AB2EZ]

Rick

Agreed! [With the caveat that the r.f. input source has a high enough impedance]

The tuned circuit between the grid and ground will provide the same self-biasing effect as would be provided by a series capacitor between the tuned circuit and the grid. The capacitor in the tuned circuit will, indeed, accumulate more charge on the negative half of each r.f. cycle, than on the positive half... because, when the grid goes positive with respect to the cathode, some of the charge that would have gone into the capacitor will, instead, go into the grid. The effect is the same... but perhaps a little more subtle. The capacitor will charge up until the current flowing through the grid resistor, in parallel with the resistance of the r.f. input source (throughout each r.f. cycle) drains off enough charge from the capacitor to remove the excess charge added each cycle when the grid conducts.

Thanks for pointing this out!

Stu

[W3RSW]

Stu,
We're both were almost there in the grid leak discussion.  For the rectification effect to work properly the time constant of the "suspect" cap. and grid leak resistor has to be considerably greater than that of the freq. of operation (TC at 1/f.)

So I suspect that the "suspect" cap. is really the RF bypass around the grid. resistor, (in the 'bottom' of the circuit and 'below' the parallel tuned circuit,) one more like 0.001uf.
Or more likely it is the combination of the 100pf or so tuning cap. and the 300pf to 1000pf , just mentioned, bypass to ground.  That gives us a time constant considerably greater than the 1/f of the op. freq.

In my literature*, I see the grid leak resistor between the coil (no cap. shown or just understood in the parallel tuned circuit, or the coil is self resonant as in early BC crystal sets) and the grid of the tube with only one capacitor connecting around (bypassing at RF) the resistor.  But the circuit is in the power amplifier, grid leak bias section and corresponds with your earlier 'word' diagram.

I'll reference exactly what is said in the McGraw Hill, and corresponds pretty much with what you said earlier with your diagram but with the additon of an increased time constant greater than that of the operating frequency's, which is key.

"A fourth method of obtaining grid bias which is often used in oscillators and rf amplifiers is shown if fig. 2.31d, This is called grid leak bias. The positive portion of the incoming signal initially raises the grid to a potential which is positive with respect to the cathode, causing grid current to flow which charges the coupling capacitor Cc. The discharge of Cc Through Rc is made very nearly constant by making the time constant CcRc long compared to the period between positive peaks of the input waveform.  When Rc is made large, sufficient bias (neg. with respect to cathode) will be developed so that grid current will flow only during the most positive peak of the incoming signal." (i.e., class C)

To clear up a little more mud, in a receiver grid leak detector circuit, the grid leak resistor bypassed by a 100pf or so capacitor, both being being between the LC resonant circuit and the tube grid is of high value, 1 or 2 megs. This naturally give a high Time constant way above that of the resonant LC circuit.
    Example. TC or CcRc for 100pf and 1 meg. = (100 * 10^-12) * (1 * 10^6) =  100 * 10^-6  = 1 * 10^-4 seconds.
      TC of a typical 1 Mhz signal is, of course, 1/f or 1* 10^-6 seconds so we see that CcRc is much (100x) greater.

In the case of the power tube circuit, your looking at grid leak of say 10k bypassed by say, .001uf. (1000pf)
      CcRc = (1000 * 10^-12) * (10 * 10^3) = 10,000 ^ 10 -9 = 1 * 10 ^-5 seconds.
This is still 10 times longer that the TC of a 1mhz signal and 20 times that of a 2 Mhz signal.  Might be an arithmitic error or two but you get the idea.

One more reason to make sure that RF bypassing of lower freq. ham band designs, 160M, is sufficient compared to most stuff you see in the handbooks designed for 80 meters and down.

***************************************************************************

*"Electronic Designers' Handbook," Landee, Davis & Albrecht, McGraw Hill, 1958, pg.2-2.7, fig. 2.31(d).
Also various ARRL handbooks of the era.

"the mores I reads, the lessens I knows"
R.

[WA1GFZ]

If I may suggest. Move the strap from the plate tuning cap to the top side of the plastic spacer of the tank coil so you have strap against strap. This way you don't have the resistance of the screw in series with the joint. This could be a real hot spot on the upper bands. I would put a nut on both sides of the plastic so you can have a tight electrical joint and a loose joint on the plastic. This way the plastic won't fracture. gfz

[W7SOE]

If I may suggest. Move the strap from the plate tuning cap to the top side of the plastic spacer of the tank coil so you have strap against strap. This way you don't have the resistance of the screw in series with the joint. This could be a real hot spot on the upper bands. I would put a nut on both sides of the plastic so you can have a tight electrical joint and a loose joint on the plastic. This way the plastic won't fracture. gfz

That is an excellent suggestion, I will make it so.

While I would like to tear open the bandswitch box, I really don't have time.  Looking inside I see no additional capacitors so I think Rick is right, the grid input capacitance combination of the grid tuning caps and the bypass caps.

I do believe the grounded pin 5's is a typo.


Last night I decided it was time to fire this thing up!  :-O

Both the plate and mod HV supplies had been brought up slowly on the bench.
I connected a my JVII to the input and a dummy load and scope probe on the output.

I left the modulator HV off, filaments on.  I fired up and tuned the JVII at 3870 Mc and I was able to peak the 813's grid current using the RF input tuning and grid tuning controls.  I set the 813 grid current at 20mA.  I adjusted this using the grid drive of the JVII, it takes VERY little to get sufficient drive, the JVII grid drive is close to zero.  Really not sure if this is the way to do it.....

The plate HV is on a variac, fired it up and put about 50mA on the 813 plates.  I was able to tune the plate tank for a minimum nicely.  There is an antenna capacity switch as well as an antenna loading control, I adjusted these for a minimum plate current, is this correct?
I then re-tuned the JVII tank.

I slowly increased the B+ until the 813 plates were drawing 300mA. 

The screen current came up to about 35-40mA.

At this point I am really sweating!  No sparks or smoke, just a lot of impressive humming.  The output looks good on the scope, a smooth sine wave, and it is bigger than the one made by the Viking!
I let it run like this for about 5 minutes and the shut it down.  My dummy load is one of those MFJ dry 300 watter's and it was beginning to smell.

I only hope this was somewhat close to proper operation.  Very exciting, weee!

Rich

[KL7OF]

The Viking 2 will provide way more drive than you need ...for the 813's class c....   20-25 ma on the grids with 400 volts on the screens is what works for me....  I drive my 813's with an 807..... What are you going to use for a speech amp?

[W7SOE]

The Viking 2 will provide way more drive than you need ...for the 813's class c....   20-25 ma on the grids with 400 volts on the screens is what works for me....  I drive my 813's with an 807..... What are you going to use for a speech amp?

I was planning on using the audio section of the JVII for the speech amp.  There is a description in the manual that shows how to break out a 500 ohm audio signal.  I know the VII audio section is poor but I also plan on doing the K6AD mods.

Rich

[John K5PRO]

Don't know if someone already mentioned it, but the blocking capacitor you have looks like one of the ones rated for DC application, and possibly 15.75 KHz horizontal oscillator application in old TVs. It has loss and inductance. If you can get one, use a 5 kV ceramic doorknob such as Jennings, High Energy, CRL, you see them at hamfests. They are only available up to about 2500 pf, which is adequate for 160 meters and up. They have wide contact on the ceramic body, with lots of solder. I measured a 1000 pF today on vector Z meter, and it stayed a capacitor (- phase angle) up to near 80 MHz. The plastic covered type that you have might have a self resonance lower, and might also add significant inductance to the circuit, not to mention being lossy enough to warm up, esp if you run low bands.

[W7SOE]

Don't know if someone already mentioned it, but the blocking capacitor you have looks like one of the ones rated for DC application, and possibly 15.75 KHz horizontal oscillator application in old TVs. It has loss and inductance. If you can get one, use a 5 kV ceramic doorknob such as Jennings, High Energy, CRL, you see them at hamfests. They are only available up to about 2500 pf, which is adequate for 160 meters and up. They have wide contact on the ceramic body, with lots of solder. I measured a 1000 pF today on vector Z meter, and it stayed a capacitor (- phase angle) up to near 80 MHz. The plastic covered type that you have might have a self resonance lower, and might also add significant inductance to the circuit, not to mention being lossy enough to warm up, esp if you run low bands.

Thank you John, I will start looking for a replacement.  How about the 580500-7 from rfparts.com?

Thanks


Rich

[WA1GFZ]

I think you want more then 5 KV for a plate blocker once you look at the plate voltage around the positive peak. 2-5KV caps in series. fc