4-1000 grid driver and MB-40 question

[KC9LKE]

In the application notes, section 2, paragraph 1, it mentions not to bypass the grid supply.
I take this to mean C15 in the 1000a schematic should not be there.

Best regards
Ted
KC9LKE

[w4bfs]

interesting thread ... found some decent reading in ARRL handbook 1956 page 153 ... could help both matching and parasite reduction

[AB2EZ]

Opcom

Although the nominal grid-to-cathode/filament capacitance of a typical tube, used in the RF output stage, may be just 10-25pF, that capacitance increases dramatically when the grid is forward biased with respect to the cathode/filament.  


[More additional charge per unit of voltage increase is stored in the volume of space between the grid and the cathode/filament when the associated diode is forward biased. The same is true for semiconductor diodes]

The effective Q of the capacitor will also drop due to the conductance associated with the flow of grid current. [The charge can more quickly leave the volume of space, where it is stored, when the diode is forward biased, and conducting forward current]

Therefore, I believe that under typical self-biased conditions (i.e. when there is grid current flowing for a portion of each cycle)... the behavior of the input circuit... particularly with respect to feedback from the plate-to-grid capacitance in series with the grid-to-cathode/filament input impedance... is more complex than it might appear based on a simple linear circuit analysis.

Therefore, a (relatively small value) series capacitor placed between the input matching circuit and the grid may have a more important role in controlling feedback/instability than first appears.

[I tried simulating the input circuit using LTSpiceIV... with the grid-to-cathode/filament of the tube represented by a 25pF capacitor in parallel with a diode; and with a 50pF capacitor between the resonant tuned input circuit and the grid. But, I didn't see the effect that I am suggesting here. I.e. at the resonant frequency of the tuned circuit, the signal produced by the RF component of the plate voltage... across the grid-to-cathode/filament input load (including the diode) ... with a 2pF plate-to-grid capacitance... was a large percentage of the amplitude of the RF component of the plate voltage... with or without the 50pF coupling capacitor in series  (i.e. too much). Therefore either the effect I am suggesting is not real, or the simulation that I am using is not completely capturing the real behavior of the circuit. Again, experimentally, adding the 47pF series capacitor between the grid tuned circuit and the grid of my 6550 plate-modulated RF amplifier produced a dramatic improvement in stability.]  

Stu


 

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.

[AB2EZ]

I think I am seeing the reason why the 47pF capacitor between the tuned grid input circuit and the grid is improving the stability.

Using an LTspice simulation:

a) With a 1000pF capacitor coupling the tuned grid input circuit to the grid, the RF portion of the plate voltage (at the input circuit resonance frequency) produces (via a 2pF plate-to-grid capacitance) an RF signal from grid-to-cathode that is a large percentage of the plate voltage amplitude, and close to 180 degrees out of phase with the RF portion of the plate voltage. Therefore one has strong feedback with close to 0 degrees of phase shift around the loop from grid-to-plate-to-grid.

b) With a 47pF capacitor coupling the tuned grid input circuit to the grid, the RF portion of the plate voltage (at the input circuit resonance frequency) produces (via a 2pF plate-to-grid capacitance) an RF signal from grid-to-cathode that is a large percentage of the plate voltage amplitude, and about 112.5 degrees out of phase with the RF portion of the plate voltage. Therefore one has strong feedback with about -67.5 degrees of phase shift around the loop from grid-to-plate-to-grid.

Stu

[Opcom]

Ted,

I have already removed that one cap and the bias is fed there through a 250 Ohm WW resistor, which I maybe ought change if it is to -not- be fed to the CT of the coil. I can add back an RFC as well.

Stu,

That makes a lot of sense to me now that it is illustrated. I have some 100pF micas (pictured) and maybe some 50pF-100pF vacuums, little ones if I can find them. What would be most suitable? I guess I have to use an RF rated one in the grid ckt. How about a little air variable rated 500V?

Xc
mhz   pf   Ohms
3.5    50   909
4   50   796
7    50   455
7.3   50   436
14    50   227
14.5    50   220
21   50   152
22    50   145
28   50   114
30    50   106

[AB2EZ]

Opcom

I did the simulation on 10 meters (28.86MHz) and the effect of the 50pF coupling capacitor appears to be even more favorable at that frequency. The phase shift around the loop is now around 180 degrees.

Notice how things change when I remove the diode (slide 7) representing the grid-to-cathode diode of the tube. There is now 360 degrees of phase shift around the loop.

So, I would suggest that you use a coupling capacitor whose value is around 50pF (twice the input capacitance of a 4-1000A)

Hopefully, this will improve the behavior of the amplifier on all HF bands

Stu

[Opcom]

I should leave my C1, the 'lower half of the neutralizing voltage divider" in the circuit? Or remove the neutralizing because of this feedback?

[AB2EZ]

Opcom


I have changed my answer several times... as I continue to ponder how this circuit works with respect neutralization (or lack thereof), etc.

Latest version: August 14, 2014...  9:45 am

First... going back to the "Colonel Turner RF Deck" schematic, and (separately) the National MB-40 Application Notes:

If there is no RF path to ground from the point where the top left variable capacitor connects to the bottom left variable capacitor of the MB-40, and also no RF path to ground from the center tap of the right side coil (i.e. with the removal of C15), then C1 provides the only RF return path to ground for the entire MB-40.

I think that with the removal of C15, you would want C1 to be a larger value than 25pF. The MB-40 will no longer be balanced with respect to RF ground if you remove C15.  Looking at the Johnson 500 schematic... the tuned circuit for the grid of the 4-400A (or for the output of the driver... depending upon how you want to think about it) returns to ground through a total of 550pF of capacitance. In the Johnson Valiant, it is 1000pF.

Alternatively, you could keep C15 (0.005uF) between: the point where the top left and top right variable capacitors of the MB-40 connect together, and ground (as per Figure 3 of the application notes). I really don't think it would make that much difference if you also keep the center tap of the coil connected to this point... as per the Colonel Tucker schematic.

If you keep C15, I suggest that you re-balance the MB-40 by adding a 50 pF series capacitor between the top of the MB-40 and the top of C1... to mirror the 50pF capacitor that is going to be added between the bottom of the MB-40 and the grid of the 4-1000A.  

I would be inclined to keep C1 to provide an RF path to ground at the top of the MB-40. It will be essential if you remove C15. If you don't remove C15, it will be useful to maintain the balance of the MB-40 (with the added 50pF capacitor between the top of the MB-40 and the top of C1); whether or not you keep neutralizing capacitor C2.  I would be inclined to keep C2 to provide some degree of neutralization. Since C2 is designed to be small (comparable to the plate-to-grid parasitic capacitance of the tube), the voltage divider formed by the combination of C2 and C1 may result in too small a signal being fed from the plate of the tube to the top of the MB-40... but I would still keep C2... and I would try to adjust its value to produce approximate neutralization.

I've never really understood why/how the method of neutralizing used in the Johnson 500, the Johnson Valiant, etc., works... but, everyone seems to use it. I'm just now (as a result of the thinking and simulation I have done for my posts in this thread) beginning to understand how it works (maybe).

 

Stu

[AB2EZ]

I added the 2nd half of the balanced input tuned circuit to the schematic... and did LTSpiceIV simulations.

Note: in the simulations below, and in my later posts, I have not included the effects of magnetic coupling between the upper and lower inductors.

Attachment figure 8: The balanced version of the input circuit

Attachment figure 9: V(n004) is the voltage source, V2; V(n003) is the voltage between the grid and the filament (i.e. across C2), and V(n005) is the voltage across C6.

Attachment figure 10: Same as in figure 9... but with the neutralizing capacitor (C7) removed.

As expected...with the center point grounded, and no magnetic coupling between the two inductors,  the 2nd half of the circuit, including the neutralizing capacitor, doesn't seem to make a great deal of difference... although it has some favorable effect. The peak amplitude of the signal produced between the grid and the filament (i.e. the voltage across C2) by the time-varying plate voltage is reduced from around 140V to around 100V.

Stu
 

[AB2EZ]

Here are the simulations for the balanced design... with the center tap lifted off of ground (C15 in the Colonel Tucker diagram is removed)... as recommended by Ted (KC2LKE). Note that C1 (in the Colonel Tucker diagram) has a value of only 25pF, and it is the only remaining RF path to ground for the MB-40.

The neutralizing capacitor (in place in attached figure 14 vs. removed in attached figure 15) now makes a bigger difference in the amplitude of the signal produced on the grid of the tube by the time-varying plate waveform... but the phase shift is unfavorable compared to the case where the center of the balanced input circuit is at RF ground.

 
Stu

[AB2EZ]

Last... but not least:

Here is a fairly conventional design that is similar to the design used in the KW-1, Johnson 500, etc.

The capacitor called C15 in the Colonel Tucker diagram is removed ... as per Ted's (KC2LKE) earlier recommendation)

The capacitor called C1 in the Colonel Tucker diagram is increased to 500pF to provide a lower impedance RF return to ground.

The 50pF capacitor between the bottom of the MB-40 and the grid of the 4-1000A is still in place.

I played around with different values for the neutralizing capacitor (C2 in the Colonel Tucker diagram). I found that as I increased the value from 0pF (neutralizing capacitor removed) to 70pF*... the voltage between the grid of the tube and the filament of the tube (i.e. across C2 in the attached figure 16), produced by the time varying plate voltage, decreased monotonically from around 42 volts peak to around 14V peak. The phase remains very favorable (i.e. in phase with the time varying plate voltage)

Attached figure 17 shows the behavior of the circuit with the neutralizing capacitor value set at 70pF*

V(n003) is the voltage from grid to filament (i.e. across C2 in attached figure 16)

Attached figure 18 shows the behavior of the circuit with the neutralizing capacitor removed.


Therefore... I think that the best configuration, so far, is this one (referring to the Colonel Tucker design):

Use a simple, unbalanced grid matching circuit configuration (no C15)
Change C1 from 25pF to a larger value...e.g. 500pF
Include the 50pF capacitor between the bottom of the MB-40 and the grid of the tube
Make the neutralizing capacitor C2 a much larger value...as needed... to achieve neutralization.

*Note: In all of my simulations, I have used a value for the plate-to-grid capacitance of the tube of 2pF. The value in the 4-1000A data sheet is specified as 0.24pF. If I adjust my simulation to use 0.24pF for the tube's plate-to-grid capacitance... then the optimal value of the neutralizing capacitor is around 10pF (not 70pF).

Stu

[Opcom]

I can see I'll be busy this Saturday.

The neutralizing cap C2 is a #10 wire through an insulator about 1" from the Eimac chimney. Its height is somewhat adjustable but is set at max now, about 5".

I would be worried that a 500pF cap from the MB-40's top (plate) end to GND would mess with it severely, but if that is the only path to ground, then the tank would basically be sitting on that cap and not influenced by it. I assume then that I should feed bias to the grid directly avoiding using the MB-40 for that as previously suggested.

What value of RF choke is acceptable for feeding the grid bias to the grid? Would a small plate choke stolen from an old 50-100W transmitter or even a junk 100W sweep tube "10M" linear work?

[AB2EZ]

I think you are going to need a 10pF (maximum) HV variable for a neutralizing capacitor.

With C15 removed, you need the 500pF capacitor to provide a reasonable impedance path for RF current to flow around. This assumes that all remaining (parasitic) paths to ground involve capacitances that are much less than 500pF.

Yes, feed the bias directly to the grid via a small plate choke, with an RF bypass capacitor to ground on the DC bias input side of the choke.

Most of all... be extremely careful!

Don't forget that the full plate voltage of the 4-1000A is going to appear on one side of the HV variable neutralizing capacitor. If it arcs over, the full plate voltage of the 4-1000A is also going to be applied to one side of each of a number of other capacitors, including one side of each of the MB-40 capacitors, as well as the right side MB-40 inductors.

You might want to add a safety choke, to ground, across the new 500pF capacitor. I.e. from the cold side of the neutralizing capacitor to ground. A few mH (milli-henries) would work... just as it does across the output port of a transmitter. The choke will also prevent a relatively small fraction (about 10/500) of the full B+ from appearing across C1 (and everything connected to the top of C1) as a result of the charge that flows from the plate to ground (via C1) through the neutralizing capacitor when the B+ is turned on.

An alternative approach, that keeps B+ off of the neutralizing capacitor, is to attach the neutralizing capacitor to the tank side of the 4-1000A plate DC blocking capacitor, instead of directly to the plate of the tube.

Stu

I can see I'll be busy this Saturday.

The neutralizing cap C2 is a #10 wire through an insulator about 1" from the Eimac chimney. Its height is somewhat adjustable but is set at max now, about 5".

I would be worried that a 500pF cap from the MB-40's top (plate) end to GND would mess with it severely, but if that is the only path to ground, then the tank would basically be sitting on that cap and not influenced by it. I assume then that I should feed bias to the grid directly avoiding using the MB-40 for that as previously suggested.

What value of RF choke is acceptable for feeding the grid bias to the grid? Would a small plate choke stolen from an old 50-100W transmitter or even a junk 100W sweep tube "10M" linear work?

[KD6VXI]

My idea,  Opcom,  was to use a small amount of L on the grid side,  in series,  to cancel out Cstray and Cin.

Similar to the L used on tanks on 10 and 6, before Ctune.

This would kill any imbalance Cstray was causing in the input.

I may be off in left field.

--Shane
KD6VXI

[Opcom]

Maybe not. The thing is there on its side and the MFJ is there, scopes, meters, all the stuff, except an easy way to run it 'live'.  I'm going to experiment a lot tomorrow. First I have to add the 2 turn coil to drive the high frequency section, if the MB-40 is to be of any use above 7.3Mhz. That is a proven thing in several articles. Then I am going to try what has been suggested to the best of my ability and take notes.

A thing that concerns me about adding inductances is the possibility of creating unexpected tuning circuits.

I already learned that an imbalance of capacitance on either the grid or neutralizing side when the MB-40 is used in a balanced manner (my original circuit) yields two resonant points instead of one. So the same could be for inductance discrepancies.  

It is also easy to un-ground the MB-40. The coupling coil being in the center of the grid coil, it may not be quite the same driving it. There is an electrostatic shield, it may help. Anyway more later. After I create a lot of unexpected things.. one of which should be a working grid section.

[Opcom]

Apologies, been real sick off some bad pizza rolls, nothing done this weekend I want to mention, At least I should be able to go to work tomorrow.

[AB2EZ]

Opcom

Feel better!

Stu

[Opcom]

here's what I have been working on based on the good advice here.

The bypasses for the tube elements can be ingored, they are just for show as I'm keeping what has been there. Please just look at the MB-40, the 100pF cap in series between the MB-40 and grid, the way the bias is put in, and the neutralizing.

I think this reflects the suggestions for floating the MB-40, having:
* the ground return for it being a 500pF cap C1 which also acts as the low-Z portion of the neutralizing divider,
* the small neutralizing cap Cn (I showed it variable)
* a protection choke across C1 the 500pF cap.
* the bias injection right to the grid
* not bypassing the bias at the MB-40, and floating the MB-40.

The transmitter also has a swamping switch in the grid, so that is now shown. I need to further redo the diagram anyway.
As mentioned, a 2T coil added to the MB-40 coupling to the high frequency coil fixed a drive issue. It is from suggestions in QST etc.


Anyway it's not quite done yet as there's some fiddling to get things right.

[AB2EZ]

Opcom

Looks good!

One suggestion. If you connect the neutralizing capacitor to the right side of C3 (as shown in your new schematic), then you might want to move the 10mH 1A safety choke (L7) to the output side of the pi network. The combination of the pi network's coil, in series with the relocated safety choke, will keep HV off of both the antenna output connector and the MB-40.

Stu

[Opcom]

some progress made and parts found.

for bias choke: a 40uH plate choke from a CB leen-yar

100pF and 400pF mica RF caps found for for the same in the diagram (400 sub for 500).

Wound a planar 2 turn coil to be fitted over the center of the high frequency coil on the MB-40. In this way it will minimize the capacitance to that coil because that HF coil has no 'cold' end, and the sections are close together. The would coil is 0.3uH and made of #14. It's held together by cable ties and some military glue. Hope that works.

[Opcom]

the pdf shows the RF arrangement now. (it is in the top left corner) Please refer to the components in the file of this post, as this is the real or 'official' schematic file.

ok well the idea shown in the schematic didn't work, that is would not tune the necessary frequencies. I don't really know why. Maybe i did it wrong.

low band top frequency is 6.4MHz (need 7.3), and high band top frequency is 25Mhz (need 29).

with the 400pF cap C35 to GND, upper freq limits are too low.

The 100pF cap C44 seems to not mess anything up.

going back to grounding the cap frame and the rest, but keeping the high freq coil for now having gone to all the trouble to get 20 and 15M. the unit should do 80/40/20/15.


This really isn't an end to it, after all, why didn't it work?  haha aha and he11s bells I'll never be on the air! The machine manitou will not come to me.

[AB2EZ]

Opcom

See first: my next post

It sounds like, in the existing circuit, there is about 30% too much inductance in each of the low band inductors: L8 and L9. This results in (roughly) a 14% reduction in the maximum ("top") frequency at which the LC circuit will resonate (for a given capacitance). I.e. 7.3MHz/6.4MHz = 1.14

If the low band "bottom" frequency is low enough, so that you can raise it by 14%... and still cover the 80m frequencies that you want to cover... you can remove (or short out with a tacked-on wire bridge) a few turns from each half of the secondary... i.e. from L8 and L9... leaving the 400pF capacitor, etc. in place. That should increase the "top" low band frequency.

Remember, the inductance is (roughly) proportional to the square of the number of turns. Therefore, to reduce the inductance by 30%, you have to remove (or short out) about 14% of the turns on each of L8 and L9

Stu

the pdf shows the RF arrangement now. (it is in the top left corner) Please refer to the components in the file of this post, as this is the real or 'official' schematic file.

ok well the idea shown in the schematic didn't work, that is would not tune the necessary frequencies. I don't really know why. Maybe i did it wrong.

low band top frequency is 6.4MHz (need 7.3), and high band top frequency is 25Mhz (need 29).

with the 400pF cap C35 to GND, upper freq limits are too low.

The 100pF cap C44 seems to not mess anything up.

going back to grounding the cap frame and the rest, but keeping the high freq coil for now having gone to all the trouble to get 20 and 15M. the unit should do 80/40/20/15.


This really isn't an end to it, after all, why didn't it work?  haha aha and he11s bells I'll never be on the air! The machine manitou will not come to me.

[AB2EZ]

Opcom

Here is something else you can try... which doesn't require any modifications to the MB-40.


Step 1: Remove the center bus that runs between: the midpoint of L8 and L9 , the midpoint of C33 and C34, and the Faraday shield. Leave the Faraday shield disconnected (floating). Do not run a bus between the midpoint of L8 and L9, and the midpoint of C33 and C34.

Try the transmitter again. If it works okay, then you are done. If not:

Step 2. Disconnect the top of L9 and disconnect the bottom of L9 (L9 is open at both ends)

Step 3. Connect the bottom of L8 to the bottom of C33 (where the bottom of L9 was previously connected)

This will cut the total inductance, currently associated with L8 and L9, in half (maybe more, depending on how much L8 and L9 are coupled). As a result, the "top" of the low band will move up by 41%. The top of the high band will also move up somewhat.

You can also do something similar to increase the top of the high band (if necessary):

Step 4 (if needed): disconnect both ends of L6 (leave L6 disconnected)
Step 5 (if needed): connect the top of C31 to the top of C34 (where the top of L6 was previously connected)

Stu

[AB2EZ]

Even better and even easier:

Just use the bottom half of the low band tuned circuit pair:

1. Leave the center bus in place

2. Leave the Faraday shield connected, as is

3. Disconnect the upper end of the upper inductor and the upper end of the upper variable capacitor from the 400 pF capacitor. If possible,  leave both of the ends disconnected, not connected together. If not possible, leave them connected to each other...and connect that junction to ground using a 27pF capacitor.

Run a wire from the 400pF capacitor to the center bus that connects the lower variable capacitor, the lower inductor, and the Faraday shield together.

What this will do is to make the tuning less sensitive to the grid input capacitance of the tube. This should increase the top frequencies that you can tune

Stu

[Opcom]

Disconnecting the buses is not so simple because one of them is the capacitor frame, and many of the other mods require removing the tuning unit again. I decided to ditch the 400pF cap as it was spoiling the high frequency end of the ranges, and leave in its place a small cap the same value as the tube's grid to GND capacitance. I grounded the frame of the unit, which means the junction of C34, C35, L8, and L9, and the shield. I left the 100pF cap in series between the tuning unit and the grid, and feed the grid bias via a choke. The result is very easy drive, 500W out with 20mA grid current and 30mA screen current, plate at 3KV/220mA - on 40M, but I have not had time to test the other bands just yet. 10M is apparently not tuneable/available but 80-15 seem to tune. Not tested live though.