4-1000 grid driver and MB-40 question

[Opcom]

The 4-1000 class C final conditions:
Eb 2500V
Ib 250mA
Eg2 200V (self modulates well on AM, from a 500V supply through a reisistor and choke)
Eg1 -200V
Po 400W (carrier)
Grid tank is an MB-40 with a 4 turn link.

On both bands below, the SWR from the TS-430 (exciter for the carrier) to the coupling link of the MB-40 is close to 1:1 only when grid current flows. I believe this is normal because with no current it had no real load. So the match seems good.

On 40M, the grid drive needed is about 15-20W and grid mA about 30
On 80M, the grid drive needed is about 50W and grid mA about 75-80

Well, I don't think the 80M setup is allowable, due to grid current. The question is, why does it need so much grid current on 80 Vs 40? Is this an MB-40 thing?

[WD5JKO]

Pat,

   I wonder what happens when you back off the drive on 80m?

Does it:

* Reduce the RF output for a given DC input?
* Does the input SWR rise as seen by the TS-430?
* Do the AM modulation peaks suffer?

Not mentioned, but I heard you on 40M talking about this, and you mentioned a ~ 100 Mhz parasitic. Is the extra drive needed to quench that parasitic out?

Jim
Wd5JKO

[WQ9E]

Pat,

What does the screen current look like on the two bands?  Do you have it much more heavily loaded on 80?

The 4-1000A operates much better with higher plate voltage, I don't think Eimac even provided "normal" operating data below 3000 volts for Class C plate modulated and your screen voltage is around half of what is recommended which will require much more excitation.

[w4bfs]

hi Pat …. just a hunch …. could be a bypass or coupling cap  that is ok for 40m but not quite snuff for 75m ...

[Opcom]

Pat,

   I wonder what happens when you back off the drive on 80m?

Does it:

* Reduce the RF output for a given DC input?
* Does the input SWR rise as seen by the TS-430?
* Do the AM modulation peaks suffer?

Not mentioned, but I heard you on 40M talking about this, and you mentioned a ~ 100 Mhz parasitic. Is the extra drive needed to quench that parasitic out?

Jim
Wd5JKO

* Reduce the RF output for a given DC input? - yes, as though the tube is not being driven hard enough, even though the meter says it is.

* Does the input SWR rise as seen by the TS-430? - yes but only when the drive is the same or less than what's needed on 40M. In that respect the behavior is 'normal'.

* Do the AM modulation peaks suffer? They suffer when the grid current shows same as 40M. Once the grid current is raised to the unusually high level, all is well. At that time, also the SWR is still good, but the drive shows 60W.

The extra drive quenches the parasitic.

[Opcom]

Pat,

What does the screen current look like on the two bands?  Do you have it much more heavily loaded on 80?

The 4-1000A operates much better with higher plate voltage, I don't think Eimac even provided "normal" operating data below 3000 volts for Class C plate modulated and your screen voltage is around half of what is recommended which will require much more excitation.

The screen current also runs higher on 80M for the same power output. What was 60-80mA on 40M becomes 80-120mA on 80M.
The screen volts with no drive, tube cut off, is about 500V. It drops to 200V when the grid is driven to make the carrier. It does self modulate very well, but I have only scoped it on 40M. It swung almost 1000V.

It had to have more C in loading to be rid of the parasitics on modulation peaks.

The schematic is attached, In AM, there is a 2K 115W resistor and a 8H choke in series between the screen supply and the screen. The resistor R16 is on the screen supply chassis. The AM/CW relay SW19 is to the right of that section. Increasing the screen volts also makes the already high grid current rise. It has some effect on output power but not so much beyond a certain point.

Also, for the loading, the RF deck schematic more clearly shows how the loading switches coarse and fine add more capacitance in steps.

[Opcom]

hi Pat …. just a hunch …. could be a bypass or coupling cap  that is ok for 40m but not quite snuff for 75m ...

That is possible. I used the best ones, BC quality ones. Perhaps a value is too low? or something is resonant? I less believe resonance because 80M is a low frequency but I can't say. What do you mean by not being up to snuff?

I am not sure what to check.

At one point I thought the grid current meter was off. For sure, the SWR on the driver is good and the collector current amps in the driver are higher, so if that SWR meter between the driver and the grid circuit says 60W, that is probably the right indication of drive.

There is something wrong on 80M, with the appearance of noise at modulation peaks, that can be reduced by overdriving the grid and carefully tweeking the plate tuning and loading.

[Opcom]

I know the volts are low for a 4-1000. I'd like to run more but am afraid of what would happen to the mod transformer if I did. It is vet old, and probably from an RCA BC rig running 3300V with 833's. 

What other inexpensive tubes would fit the 4-1000's place, re-use its cooling, and run well on 3000V? or even 2400V? If the HV supply was changed to choke input, it would be well regulated but have 2400V max. The modulators being 3-500Zs would be OK at 2400V, I think.



There is no shortage of power at the present voltage but if I wanted to increase it:

The transformers are 4800VCT and rated for the CT to float, being 2.5KVA pole pig types. I thought about adding a 1000V supply between the CTs and GND.

It is possible to put bridge rectifiers in and run 5000V, but would have to redo the whole supply, including higher voltage caps, and then worry about the chokes and the rest, as the chokes are rated 5KV. It's probably asking for trouble. Tucker had it set up that way and eventually blew up the power supply components, spectacularly, according to his son.

[N2DTS]

200 volts on the screen under load???

Its a triode!

[VE3AJM]

Looking at the specs for the National MB-40SL, it looks marginal for use in the grid circuit of this transmitter. Its rated for max. 20 watts in a grid input circuit and 40 watts in a final tank circuit. Perhaps it will work at higher power levels or not. And then theres the issue of the parasitic.

The schematic shows the MB-150 tank being used originally in this transmitter. A much heftier rated device. I know they aren't as common to find out there these days.

Anyone tried a passive grid input circuit to the grid of a tetrode such as the 4-1000. It would require a lot more driver power be available, but might clear up any instability there.

Al VE3AJM

[WA5VGO]

What other inexpensive tubes would fit the 4-1000's place, re-use its cooling, and run well on 3000V? or even 2400V?

4-250 or 4-400?

Darrell

[W2VW]

I've used the 4-1000 at 2750 and 5500 volts plate modulated and as a linear AMplifier.

They do have more gain at higher plate potentials but work just fine at 2750.

80 mills of grid I will damage the grid in short order as you suspect.

It's likely the drive required to get the power out is because the loading has been reduced too far to get rid of the parasitic.

Looking at the schematic:

The cathode return should be changed to a center tap on the filament xfmr. Otherwise there will be 60 hz intermodulation on the signal.

The screen bypass is curious to me. Normally it is returned to chassis ground with short large conductors. Any inductance in these connections adds to the inductance of the screen connections within the tube and can form a tuned circuit which will support parasitic oscillation.

If it were my transmitter I would calculate the largest screen bypass capacitor possible before rolling off high frequency audio and use multiple paralleled low inductance capacitors connected directly to the chassis instead of returning to the cathode.

The size of capacitor returning to the grid tank is very critical. A good indication of proper neutralization is to see grid current peak slightly or at least be almost flat while tuning across the plate dip. Grid current rising on one side of the plate dip indicates the neutralization bridge circuit is not nulling. A small change might be in order. Good luck getting it to be perfect across all bands. Better can be the enemy of OK here but you are still looking for parasitics and changes can sometimes help find them.


I'd also reduce plate loading capacitance in order to increase plate load on 80.

Fun times. Good luck.

[Opcom]

So 2000pF total capacitance may be too small for the screen grid?

The pic shows the under-socket area.  I could lift the filament and ground the CT near the transformer. The only difference being the need to add another wire from the filament voltmeter to the filament.

Would the strap between the screen terminals be sufficient? The hump there is possibly for movement, as the socket pins are free to move a little. I could eliminate it or short across it if necessary. Just the way it was built.

The wire from the screen there is just going to the bulkhead connector. It's about 10-12" long.

The 56 Ohm resistor shown at the control grid terminal is no longer present. it's just a wire now.

In the MB-40 grid circuit picture, the grid lead comes through the top of the box through a ceramic insulator. The ground lead to the capacitor frame is now a much shorter 1/4" copper strap going to the back of the box.

[Opcom]

On 80M, the loading to be rid of the parasite does not comply with the tuning charts. It's probably 900pF.

.............

The plate tuning cap is 0-300pF max cap os 00 on turns counter.

The plate roller is 0-22.5uH and 27.4 turns. max inductance is at minimum turns, so the sense is reversed.

The loading gives 0-1600pF in 100pF increments.

[[[[[[[[[[[[

The original tuning charts, incomplete, attached to the front of the unit say:


10M
final loading C not stated
final inductance turns not stated, inductance not stated
plate tuning cap not stated

11M
final loading C3 (c8+c9+c10 in the RF deck GIF file= 300pF total)
final inductance 25.3 turns, inductance not stated
plate tuning cap 9pF

15M
final loading C2 (c8+c9 in the RF deck GIF file= 200pF total)
final inductance 23.74 turns, inductance not stated
plate tuning cap 10 PF

20M
final loading C2 (c8+c9 in the RF deck GIF file= 200pF total)
final inductance 20.88 turns, inductance not stated
plate tuning cap 16 pF

40M
final loading C3, C5 (c8+c9+c10 in the RF deck GIF file= 550pF total)
final inductance 18.67 turns, inductance not stated
plate tuning cap 100pF

80M
final loading C not stated
final inductance 15.03 turns, inductance not stated
plate tuning cap not stated

[W2VW]

The grid resistor might help quench parasitics. Maybe that's why it was there.

There should be some in the plate leads of the modulator tubes also.

Each screen pin can be bypassed to ground with direct low inductance connections. A pair of doorknobs plumbed there with very short, wide copper should at least make the parasitic change it's tune.

[Opcom]

I'm reviving this because I have been working on the MB-40 grid circuit. It seems to be the cause of several of the problems.

I have not been using 80M due to antenna issues but the parasite on 80M did go away when the transmitter was heavily loaded, I thought too heavily, because I am used to loading linear amps. However, the TX needed more. I found that the loading needed to be increased a bit past the point of maximum output, resulting in more plate current. The 4-1000 does not care about this at 2500-3000V and barely shows color.

Temporarily moving the filament GND to the CT did reduce the 60Hz hum. It is very quiet in that connection, and I may make that permanent later after I get some more copper strap for re-doing the screen bypasses which are intimately involved with the grounded filament pin at this time. The hum from 'filament modulation' however is not the most objectionable thing and is very slight, maybe 5% of the total carrier amplitude. I doubt anyone hears it.

Back to the MB-40. It is proving to be troublesome to get it to work on 20M and up without messing up its tuning at 7000-7300KC. 

I added a 2 turn link around the high frequency coil as suggested in the old volumes, and sure enough, it can be driven well by that link on 20-15-10.

On the 40M band, the capacitor is now completely un-meshed at 7000-7300 KC. I had not noticed this, it was very easy to drive despite this.

I can see that the actual resonance is a little higher, like 7800KC, with the cap un-meshed. This is possibly part of the overall parasite/stability issue. The unit is not at resonance. OK well the cap is at minimum, the resonant frequency with the tube installed is too high and that's it. I am not going to start adding or removing turns from this classic part. There's no place to add a turn anyway.

This sounds counter intuitive, yes. Why would the resonance with C un-meshed be too high at 7800KC, and why would adding C by meshing the cap a little not bring it down to the 7000-7300 KC range? some hokey pokey going on.

I believe it is time to make up a good old fashioned LC circuit and band-switch, and give up the one-knob tuning on the grid. More switches and knobs is better, right?

It is not a condemnation of the MB-40. Maybe it just won't work as a grid network in this transmitter. It is not a condemnation of multiband tuners that are designed specifically for an application or for the MB-40 and MB-150 as antenna matchers (considering their stock 300 ohm "low side" coupling link).

The TX originally had a 5 band turret covering 80-40-20-15-10 meters. It was removed because it had been hacked up to add 2726KC and make the 20m coil also do 15M, and ruined by the plastic supports rotting off, but each coil had a nice coupling link and there was a switch , a very nice arrangement. I still have the original split stator cap.

Is there anywhere today I can find such a 5-band assembly in good working and mechanical condition?
Who even makes that kind of stuff any more? I am shamefully not accustomed to making my own precision coils.

Would it be as well to adapt the pi-network or just the coil from a 100W pi-network from a scrapped radio?

[WD5JKO]

It sounds like you want to move beyond the MB40, but in case you want a NOS MB-40, here is one:

http://amfone.net/Amforum/index.php?topic=36613.0

Or perhaps keep that new link for 20-15-10m, and then add a few turns on the other coil to get 40m back. It could be a moving target, but then again you might just "nail it" such that it works to your expectations.

Jim
Wd5JKO

[Opcom]

I'm continuing again, due to some more work and discovering some results.

I have not been using 80M due to antenna issues but the parasite on 80M did go away when the transmitter was heavily loaded color.

So this issue is closed due to good advice here. Thanks!

Temporarily moving the filament GND to the CT did reduce the 60Hz hum.

For now I have left this alone, no one said they hear any hum. Maybe later on principle.

Some corrections about the MB-40 and its top frequency of resonance on the low frequency side.

with tube in: 7200KC
with tube out: 7500KC
with C1, the 25pF bottom end of the neutralizing voltage divider out, 7500KC.
with tube and C1 out -not checked.

The MB-40 does not reach its highest frequency with so much capacitance loading it.

There may be enough capacitance to GND  arond that MB-40 to do away with C1. I ran a pair of 6146's with 'too much' neutralization (very small C1) and some grid swamping (5K resistor) and  it worked great, somewhat less gain, no parasitics. Way different tube, yes, but maybe it's worth a try.

As for the poor high frequency operation, the previous older experimenters' articles about adding 1-2 turns around the high frequency (small) coils and driving that, has tested very well up to 10M.  A plastic pill bottle with a slot up one side to pass the wires of the small coil will fit over that coil This slotted pill bottle should support a properly made 2 turn coil of maybe #20 to #14 wire. The new coil should be aligned over the middle of the small coil.

The input coil for the MB-40 has a Faraday shield. Should I try to make one for the new coil too?

This is a long work and a long experiment so thanks for hanging in there with me while these things are tried.

[KD6VXI]

How about measure Cstray on input and wind some Xl to buck it.

It's your stray capacitance killing you.

--Shane
KD6VXI

[Opcom]

That's a very interesting suggestion. I don't understand where the L would go because it's removing the grid capacitance and the C1 capacitance that gets rid of the trouble. Do you mean like the attached picture?

[W3GMS]

You may want to check with Bob, W2ICQ concerning the MB-40.  I know several years back he built a final using a single 4-250 modulated with a pair of 810's and it works very well all the way up to 10 meters.  I never heard him mention any problem in using the MB-40.  I myself have never used that network but he has had very good luck with it. 

I believe his email address is on QRZ.com.

Good luck with the project.

73,
Joe, W3GMS

[W2VW]

Years ago I dumped a pre-fab bridge neutralization unit in favor of a homebrewed one in order to get on 160.

Old marine medium wave transceivers have plenty of coil stock for this. Probably better to wind the sections for higher than 40 though.

[AB2EZ]

I think that you may see a significant improvement in the behavior of this circuit if you do both of the following:

A. Rearrange to grid bias circuitry to directly connect to the grid of the 4-1000A (not via the center tap of the input matching circuit)

B. Insert a 100pF capacitor between the input matching circuit and the grid of the 4-1000A

It think this would help, based on the following:

If you look at schematics for transmitters with grid-driven final amplifier tubes... like the KW-1 and the Johnson Valiant... you will usually see a capacitor with a relatively small value of capacitance between the output stage driver and the grid(s) of the final tube(s).

In a grid-driven RF output stage, one clearly needs a capacitor to block the DC (if not inductively coupled) from the driver stage, and to produce self biasing of the grid(s) of the final tube(s). But it puzzled me as to why the value of the capacitor was so low in many grid-driven transmitters. E.g. a pair of 25pF capacitors in parallel in the case of the Johnson Valiant; and a 150pF capacitor in the case of the KW-1.

Recently, I built a small, plate modulated, 40m transmitter... with a grid-driven 6550 as a final RF tube. It is a toy compared to a 4-1000A transmitter... but I did notice something that I found interesting:

I put a tuned circuit on the input side of the 6550 to obtain sufficient grid drive from the 6CL6 driver stage to drive the 6550 into class C.

However, this resulted in instability of the 6550 stage. I assume that this was caused by the high impedance between the 6550's grid and cathode... resulting from the input tuned circuit... which allowed the RF component of the plate voltage of the 6550 to couple strongly into the grid... via the plate-grid capacitance.

I added a 47pF capacitor in series (instead of the 1000pF capacitor I was using before), between the top of the input tuned circuit and the grid of the 6550... and the problem completely disappeared. For example, using my oscilloscope, I could not see any residual modulation of the 6550's grid-to-cathode (ground) voltage when I plate modulated the 6550. [Note: This does result in a reduction in the level of the 6550's grid-to-cathode RF voltage... produced by the driver... because the 47pF capacitor forms a voltage divider with the input capacitance of the 6550.]

I thought about this... but I could not come up with a simple explanation based on linear circuit analysis. Somehow, the RF impedance seen from grid-to-cathode (of the 6550) was relatively low (thus dramatically reducing the effect of the plate-to-grid capacitance)... even though the 6CL6 driver was looking into a resonant circuit...coupled to the 6550's grid via the 47pF capacitor.

I think that this is a non-linear effect that results from the fact that the grid-to-cathode input of the 6550 (or other grid-driven RF output tube) looks like a diode... and somehow this combines with the partial isolation provided by the 47pF series capacitor.

Stu

 

[w1vtp]

would this be helpful?

[Opcom]

I have the manual, however it looks like the capacitor front shaft is grounded through the dial on mine because I must have overlooked that, time to find an insulated coupling and re-check everything. Looking at the KW-1 diagram, the blocking cap is there. Its reactance would be maybe low compared to grid impedance. I could see where too large of one might have other parasitic values as well. It's worth it to try different approaches.