Schematic Check Print 813 Deck

[W2XR]

If this amplifier is to be run in class C plate modulated service, you do not need or want the VR tube to regulate the negative grid bias to the 813 grid.

In a plate modulated amplifier, you really want the grid bias to be completely unregulated; this serves to improve the linearity of the modulation process in the final amplifier circuit. The only purpose of the fixed bias supply is to provide some degree of protective/safety bias, and to support the bulk of the grid bias which should come from the grid leak resistor.

A good rule of thumb is that approximately one-third of the grid bias voltage should come from the fixed bias supply, with the balance derived from the grid leak resistor.

And I agree; why run one 813, when you can run two of them instead? 813s are still inexpensive and readily available, as are their sockets.

73,

Bruce

[W2PFY]

I think he was testing us? Did we pass

[W3GMS]

Mr. HTV (I am sorry, I do not know your real name )

First congrats for taking the incentive to build a home brew transmitter.  When you put a schematic on here you have some of the best minds looking it over for error's and suggestions. 

A couple of things I noticed in the postings:

Some folks that posted to you said the same thing that others before them have said.  Some of the comments were simply repeats. 

Don't be discourage by the comments that you made so many mistakes and the inference that you should already know better.  I take it that your new to home brewing these type of circuits and making mistakes is part of the learning curve.  The technical comments were right on, but rather blunt. 

Oh and finally, I believe you answered very well why you were not running a pair of 813's was due to a modulation transformer limitation.  That's smart planning to allow for another jug to be put in later when you find suitable iron. 

A very good reference design is from Tom, K1JJ although with the good technical comments here, your on your way to making a working amplifier.

Have fun!

Joe, W3GMS     

[KB7HTV]

Some corrections, rev1 schematic


HTV

[Ralph W3GL]

  You forgot the cap to resonate the grid coil...

[KB7HTV]

Ralph, true enough, Will add correction  to fix,
How do I set the idle current in the modulator, resistor not shown on modulator

HTV

[Ralph W3GL]

  If you are going with the typical pair of 811A's, they run 0 bias up
  to about 1250 VDC on the plates.  I've run them around 2 KV with
  a 4.5 volt dry battery in the CT of the driver xformer...

[KC2ZFA]

here's a handbook design (from '57).

[KA2DZT]

Greatly improved schematic.   I still see some problems with your design layout.

The bias supply is not correct the way you have it.  First, it is going to put hum on your grid.  I still think you should not use the bias supply and protect the 813 with a clamp tube circuit.  But if you want to use it you need to change the design. 

From your diode connect a resistor about 1000ohms, then connect the negative terminal of your filter cap 20-50mfd 200V to the 1K.  DO NOT ground the winding on the xfmr, return that side of the winding to the positive terminal of your filter cap.  You now have a half wave resistor input bias supply that is still not grounded, it is floating.  Now connect an adjustable resistor, about 5K, across the filter cap.  Connect the junction of the negative side of the filter cap and 5K to your grid circuit and then ground the adjustable arm of the 5K resistor.  This circuit will have proper filtering on the bias supply regardless of where you adjust the bias.  The bias is adjustable from zero bias (when the ground point is at the negative side of the cap) to full bias voltage (when the ground point is at the positive side of the filter cap) or anything in between.

The 2.5mh RFC in the screen circuit is not needed and can cause problems because of the (same) 2.5mh RFC in the grid circuit, may cause the 813 to oscillate.  Replace the RFC in the screen circuit with a 1-2K wire wound 10W resistor.  Put the screen bypass cap at the tube pin not where you have it.

W3GL already mentioned the missing grid tuning cap (usually about 100pfd).

The grid bypass cap, at the junction of the grid coil and grid RFC, is too large,  about 250-300pfd is more correct if you expect to get any feedback voltage from the neutralizing cap.

Add a filament bypass cap on the center tap of the xfmr.  For the filament bypass caps you can use .01mfd disc ceramics.

You'll need some way to put the 813 in low power to tune.  We can get to that in another post.  Not hard to set up.

Good luck with the project,  myself and many others are more than happy to help you through to a successful project.

Fred

[The Slab Bacon]

That one looks MUCH better! ! !

I agree with Ralph, you need a cap to resonate the grid input coil.
You also need to relocate the cap at the low side of the biass adjusting pot. The low side of the pot should be grounded. Also add an electrolytic cap on the outpoot of the bias supply for filtering. (where the diode meets the pot.)
Hook the bias supply to the high side of the pot (so the pot acts as a bleeder.)
then hook the bias feed to the grid to the wiper of the pot.

Screw the clamp tube!! It's not worth the bother if you got it biased properly.

And like Fred said, remove the RFC fron the screen feed and run it straight to the screen with a .001 bypass cap.

[WD5JKO]

Screw the clamp tube!! It's not worth the bother if you got it biased properly.

   As shown, when keyed, and with no excitation, the 813 screen voltage goes to 1500v. Is this of any concern with an 813? I doubt if the protective bias will be enough to keep the static plate current within a reasonable value.

  Also, if plate modulated, don't we want grid bias more than fixed bias?

Jim
WD5JKO

[The Slab Bacon]

Screw the clamp tube!! It's not worth the bother if you got it biased properly.

   As shown, when keyed, and with no excitation, the 813 screen voltage goes to 1500v. Is this of any concern with an 813? I doubt if the protective bias will be enough to keep the static plate current within a reasonable value.

  Also, if plate modulated, don't we want grid bias more than fixed bias?

Jim
WD5JKO

Jim,
     You're right I didnt catch that, there is no grid leak. that is an often debated subject,
whether or not to use a grid leak on a plate modded final. With a grid leak, however he will have to drive it into class c operation.

As far as the screen goes, in standby that shouldn't matter in his circuit. With the filament center tap opened up, there is no DC path to ground for it to draw any screen current, so it shouldnt matter as long as the voltage isn't high enough for it to arc over to another element in the tube.

Back to the clamp tube for a second, it should only be necessary if you are only using grid leak biass. The fixed biass should be set at a level that would protect the tube in the event of loss of drive.

[WD5JKO]

Back to the clamp tube for a second, it should only be necessary if you are only using grid leak biass. The fixed biass should be set at a level that would protect the tube in the event of loss of drive.

  Yes I agree, but as G2 rises towards 1500v in the case of lost excitation when rig is keyed, the -100v or so fixed bias will not be enough to keep the tube from overloading. On a tetrode, the G2 voltage has a very pronounced effect on the G1 to Ip characteristic.

  Isn't it more common with big tetrodes, and beam power tubes to get the G2 voltage from a different source than the plate voltage? I'm thinking the Globe King 500 where the screen gets a high impedance 500v G2 supply, and with a choke it self modulates G2 along with plate modulation, or the ART13 where we have separate supplies, and where the modulation transformer has a separate screen winding.

Just my 2 cents..
Jim
WD5JKO

[The Slab Bacon]

Jim,
      If the fixed biass is set to a point that would cut the tube off with the screen voltage and plate voltage appiled, it shouldn't matter much. However, I do agree that a tertiary winding on the mod tranny or a self modulation choke would be a better way to fly.

Since he allready has an ART-13 mod tranny this would be an easy way to go.

He could also tap down the HV on a voltage divider instead of just using a dropping resistor.

If he was to change the biass rectifier to a simple half wave doubler this would give him all the biass he could ever need.

This is kinda fun, all the fun of doing the design work without having to do the drilling and blasting.............................

[WD8BIL]

Isn't it more common with big tetrodes, and beam power tubes to get the G2 voltage from a different source than the plate voltage?

Although it is done that way, Jim, I don't know ifn it's more common. Having the scrren and plate come on together limits the chance of the screen being on without plateV which, of course, usually will only happen once! 

[KA2DZT]

I would get rid of the bias supply and use the clamp tube.  I have a clamp tube in my 813 and it worked perfect.  I shut off the drive (with full plate voltage) and the screen voltage dropped a very low value, plate current also dropped to a low level.  It's better to get the grid voltage from a grid leak resistor.  I used 13K about 10w.  At 13ma of grid drive you get about 160-170 volts on the grid which is the recommended amount.

I also have a tune position which is controled by the screen voltage.  I switch in a 7.5K 25w resistor from the screen side of the 40K screen resistor to ground.  This drags down the screen voltage but still leaves enough plate current to easily tune the grid and plate.  The OP is using a screen current meter shunt resistor, so, the 7.5K 25w resistor should be switch in at the 40K but before the screen shunt.  Same thing is needed if he uses the clamp tube circuit.

I also think the plate tuning cap is too large.  The drawing shows a 500pfd.  About 200pfd is more than enough to tune on 80M with his tank coil of 19uh.  You need a high voltage tuning cap.  I use a 80pfd tuning cap (9KV 1/4" spacing) and a 118pfd 15KV RF doorknob on 80M, but a 7.5KV cap would also be enough.  I mention this because of the large size of these caps.  Also with a 500pfd cap it makes tuning very touchy.  Using a vacuum cap is always better.

IIRC from I what saw on the drawing, the plate blocking cap of 5KV should be increased to at least 10KV, I'm using two 510pfd 18KV doorknobs in parallel.  The plate bypass at the plate RFC should also be at least 10KV .001pfd.

The loading cap of 1000pfd probably is just enough for 80M.  Any low voltage type tuning cap can be used.  Usually most often used are the BC type tuning caps.  I use a 4 gang 2100pfd on 80M, 2 gangs on 40M.  Not sure how much of it is in use.  I think about half meshed on both bands.

Fred

[KA2DZT]

Isn't it more common with big tetrodes, and beam power tubes to get the G2 voltage from a different source than the plate voltage?

Although it is done that way, Jim, I don't know ifn it's more common. Having the scrren and plate come on together limits the chance of the screen being on without plateV which, of course, usually will only happen once! 

Thanks Budly,

A very often overlooked problem.  That's why it's better to get the 813 screen voltage through the dropping resistor from the HV supply.

Even with my 6550's that I use for the 6146 the 650V is always on the plates of the 6550's even on standby.  The 300v screen voltage is on the PTT line.  But, there is a safety relay that will open the screen circuit (within 1 sec) if the plate voltage is removed.

Fred

[KB7HTV]

Hello all,

Made the suggested change on the grid bias supply, should have 80 ma. max. in the floating circuit , the tube table suggests -90 volts here, some of you suggest this be higher. the limit would be the voltage of the secondary of the bias transformer, should I go ahead and add a series resistor to increase this.
the tube table lists a grid current of 12ma. can this listed value be used to size the grid resistor? if so what total bias should I be shooting for then I can back out the resistor size. ( what about the 1/3 rule?)

The air caps I have for the tank circuit are of an unknown value, and I have additional air caps in the junkbox, I was shooting for the values on the schematic. the plan was to use old russian doorknob caps to increase the value of the air variables if needed, thinking the tune range would not have to be all that wide for this thing to live at 3870kc and never move I have several air caps with almost 1/4" spacing, I can sort of guess at thier value comparing them to ones of known value.

here's rev 3
will a broadcast cap live in the load position, I don't know

[KB7HTV]

HV supply

Sizing the HV Bleed resistor:

Unknown : HV transformer about 40 pounds worth no output capacity in ma. data printed on it

using 1500-0-1500 taps on it has also 2000-0-2000 taps.

Known: Full wave rectifier with Big crazy Silicon valley rectifiers Full wave not Bridge.
they are sized at 5KV each for 2 amps. (I have 4 of them using only 2)

Mombo unknown value HV choke about 30 pounds, after the rectifiers in the HV line, followed by a stack of 47Uf electrolytics 8 I think 450V each.

with a resistor stack across the top (25k 15watt ea.) (from cap input arrl supply)

before adding the choke (Cap input supply) the thing climbed to 3200kv (no load) and I shut it off with a stick on the breaker
Need to size bleed resistor, and add the choke to stop this. do I do this by trial and error not knowing what output the transformer is good for or say use a 15k 150w resistor an pull 100ma. right off the top ans see what it does.
if I just add the choke make it a Choke input supply what will it do, the caps won't see the rectifier ripple and it may not pump up to the ac peak.
or will it?
HTV

[The Slab Bacon]

Look at your grid circuit / biass supply closely in rev3.pdf. You still dont have a grid leak or a DC path to ground for the grid / biass supply.


Fred,
       You ALWAYS take the long way around the block. Do you also walk around with a shirt that says "Kick Me" on the back! !  ! ! !      Lets face it, a 3' tall Budd cabinet stuffed to the gills for a 50w transmitter is a bit of overkill. 

[W8ACR]

Why don't you really simplify things by using an 810 instead of an 813. Get rid of all that screen circuitry mumbo jumbo. Use the 813's as triode connected modulators. To quote K9ACT...... "triodes are nice"

Ron

[KA2DZT]

I haven't look at the rev drawings but just to answer a few questions.

BC radio tuning caps are fine for the loading caps, you made need a three gang one if you have one,  if not, use a two gang.

Finally you told us what freqs you are building the rig for.  With that 19uh coil you'll need less than 200pfd to tune the tank circuit on 75M.  Try to use a 100pfd wide spaced tuning cap and another 100pfd, or something close, HV RF doorknob, if you have one.

I'll have to re-read your antics with the power supply testing.  Try not to electro-plate yourself.  FORGET the capacitor input supply.  That type of supply is far too unstable for what you're building.  Especially with the high-bias-cathode-resistor-relay scheme you're using for standby.  Use the choke input filter.

The 25K 15 watt resistors that you're placing across each of the eight filter caps will give about 10ma of bleeder current at 2KV standby voltage.  You can reduce the 25K's to something less 15-20K if you have them.  You will need to increase the bleeder current to at least 30-40ma total.  So you will need to add another bleeder group of resistors across the output of the supply to make up the difference from whatever the equalizing cap resistors are drawing.  I usually use two large power resistors in series for bleeders.  With the eight 450V caps in series the filter cap is good to 3.6KV.  No need to get close to the 3.6KV on standby, when the supply is lightly loaded.  30-40ma of bleeder current will keep you below 3.6KV even with the higher voltage taps on the xfmr (with the choke input filter).

With the grid bias supply you will need to add a grid leak resistor.  The grid leak resistor will need to drop the grid voltage needed ( with 13ma of grid current) minus the grid bias supply voltage.  The total grid voltage needed is about 160-170V for the 813 at about 1500 volts on the plate.

Fred

[KA2DZT]

Are the filter caps 47uf or 470uf??  You said they are 47uf.  If so, you will not have enough filter capacitance with eight of them in series.

Let me know.

[KA2DZT]

Took a close look at the two schematics.

I see you change and added all the improvements.  There is one thing you didn't get exactly correct.  You connected the grid lead to the wiper arm on the 5K resistor.  There is no ground point.  I said to connect the grid lead to the junction of the 1K, 5K, neg terminal of the filter cap and ground the wiper arm.  You can also do it the way you have it, but you then must ground the junction of the positive side of the filter cap, 5K and xfmr lead.  Either way will work.  You will also need to add in a grid leak resistor between the bias supply and the meter shunt resistor.

It's now 5AM, so will continue later today.

Your metering circuits will need some changes but will get to that in another post.

By the time you build this xmtr you'll either be an expert or as confused as all of us are

Fred

[KA2DZT]

Look at your grid circuit / biass supply closely in rev3.pdf. You still dont have a grid leak or a DC path to ground for the grid / biass supply.


Fred,
       You ALWAYS take the long way around the block. Do you also walk around with a shirt that says "Kick Me" on the back! !  ! ! !      Lets face it, a 3' tall Budd cabinet stuffed to the gills for a 50w transmitter is a bit of overkill. 

Frank,

Here we go again, lets pick on Fred. (good title for a net)

The 3ft 18" deep open cube rack contains the 813, 811's, 1500V PS, 20amp variac, mod iron, 50hy mod reactor, and the whole 50watt xmtr.

I was just reading one of your other posts.  You bring up a good point.  The very high screen voltage that occurs  when you open the filament CT on standby. Without any screen current the screen voltage could rise to a the full unloaded power supply voltage, which may reach 2500V.  May arc over inside the tube.

Fred