The AM Forum
March 29, 2024, 10:36:05 AM *
Welcome, Guest. Please login or register.

Login with username, password and session length
 
   Home   Help Calendar Links Staff List Gallery Login Register  
Pages: 1 2 [3] 4 ... 6   Go Down
  Print  
Author Topic: Simple Audio Driver for 833A modulator  (Read 94354 times)
0 Members and 1 Guest are viewing this topic.
nu2b
Guest
« Reply #50 on: November 03, 2005, 08:30:22 AM »

Frank,
I seems possible that if the gate is driven to ground that the tube can
pump enough current through the effectively open ckt (high Z) fet drain,
even if its only 10 mils, such that the fet breakdown voltage will be exceeded.
It's very hard to read the tube curves here.

You mentioned a resistor to ground. The nice thing about the 2 resistor,
1 FET KISS circuit is that along with the local negative feedback to improve the distortion,
stabilized dc gain, (and fixed ac gain) is that you already have a built in resistor to gnd.
Note that if the audio driver to the gate has too low an output impedance, it will negate
the local negative feed back and you will need to wrap neg feedback around the driver.

Regards,
BobbyT
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #51 on: November 03, 2005, 09:24:38 AM »

Hi,
Yup phase splitter Z could be an issue. You want the drive Z low enough to avoid RFI problems but not too low. Still injecting feedback is no big deal. I think someone should try my circuit then if there is problems remove the phase splitter and add KLR op amp drive on a small board. The op amp interface to high C gates may need attention. OP amps don't like high C loads and tend to oscillate. This can be eliminated with a small chance to the feedback configuration. Values will depend on the op amp and load.  A 50K resistor across each driver would keep the voltage from soaring above the rating of the FET. Yes I don't have any clue of tube curves at very low cathode currents. I know from GG linear days the 50 Kohm load across the cathode relay contacts limits the voltage. So what if the tube leaks a couple mills it will be more stable rather than sitting as a hair trigger wanting to oscillate.
I take it you agree we don't need a pullup but a 50 K pull down is good insurance.
Logged
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


"Let's go kayaking, Tommy!" - Yaz


« Reply #52 on: November 03, 2005, 10:46:47 AM »

Frank,
Sounds like we're getting closer to building a prototype. 

Mike/ZE said he wants to build something this weekend for his rig.  So maybe the next move might be to simulate it the best you can and come up with a first schematic to try, based upon the current thought.  Mike's tests will give us a lot of data to work from. Give Mike a list of test parameters you want checked.

If Mickey builds it with some room to spare to add to/ modify later, it might work out FB. I won't be building mine for at least a month. I just built two chassis for the 833A final and modulator last night, so far away yet.

T

Logged

Use an "AM Courtesy Filter" to limit transmit audio bandwidth  +-4.5 KHz, +-6.0 KHz or +-8.0 KHz when needed.  Easily done in DSP.

Wise Words : "I'm as old as I've ever been... and I'm as young as I'll ever be."

There's nothing like an old dog.
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #53 on: November 03, 2005, 11:00:25 AM »

Tom I would start with the schematic I sent you and add 2 50K resistors drain to ground. This will add some protection to the FET as BobbyT mentioned when the tube that is off swings above 6 KV. My bias circuit should work. If it doesn't then KLR circuit will need temperature compensation. Just use regulated power supply for the driver. Use a heat sink for at least 100 watts dissipation. Then leave open space to add a little board later if necessary. Sorry don't have the time this week to play simulation. Work gets in the way. You retired guys can do that.  fc
Logged
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


"Let's go kayaking, Tommy!" - Yaz


« Reply #54 on: November 03, 2005, 11:30:35 AM »

OK, I made changes on the schematic.

Hopefully Mike will try it.

With the schematic as shown, it appears more gain is needed to go from 1/2V  line level input, right?  Maybe this is where an op amp and some neg feedback will be introduced later on.

The cathode fets are the 900V 10A ones, but what do we use for the phase splitter fet... will one of those work there too?

T
Logged

Use an "AM Courtesy Filter" to limit transmit audio bandwidth  +-4.5 KHz, +-6.0 KHz or +-8.0 KHz when needed.  Easily done in DSP.

Wise Words : "I'm as old as I've ever been... and I'm as young as I'll ever be."

There's nothing like an old dog.
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #55 on: November 03, 2005, 11:46:34 AM »

Tom,
I would also use a high voltage fet for the phase splitter just in case a transient is coupled back. Don't forget transorbs gate- source. coupling cap values will effect low end response. I suspect .1 to 1 UF film will work. Radio shack had a good selection of them. Also put a .001 to ground on each filament lead to keep the tubes from oscillating. A .01 ceramic + 1 to 10 UF on each bias pot wiper would also be a good idea. Remember to have bias pots set to zero when you first key the rig so the tubes don't get red in the face. This circuit could be tested in any rig with 2 filament transformers in the modulator. The circuit will work a lot better if all the resistors are noninductive. but it should function with wire wound. This is only to keep any inductance out of the cathode setting up a tank circuit.
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #56 on: November 03, 2005, 11:49:44 AM »

BTW you will need more than line level but another stage could be added later.
FET would need 4 resistors and a cap. Same power supply. or an op amp stage would also work.
Logged
Tom WA3KLR
Contributing
Member
*
Offline Offline

Posts: 2120



« Reply #57 on: November 03, 2005, 11:53:52 AM »

Some status -
Well, I’ve simulated my 1st-pass design and Frank’s.  By the way, I realized that I left out 47k series resistors ahead of U2 and U6’s (-) inputs.  Both circuits do appear to work.  I simulate the voltage and current load that the 833 may present to the FET drivers by using a 125 Ohm pull-up resistor to +100 Volts.  I am not simulating the full application unfortunately!

I found a triode, tetrode and pentode in the SwitcherCAD library and I used the triode symbol in my first published schematic that came via Tom K1JJ.  But later investigation reveals that they are symbols only with no macro-models.  I am not going to attempt to develop models for the tubes.

Since we don’t have the ability to simulate 833’s, the sine output of the modulator transformer isn’t available.  I can’t analyze the class B waveform, so I have to keep the drive level down to the point where the cathode signal is still class A.  I am seeing about 2.0% THD on my circuit and about 4.6% with Frank’s.  The distortion would increase at higher drive levels that will actually be required.

One problem I see with my circuit is that the bias current levels rising with drive.  Frank’s circuit, the bias point remains at 50 mA. per tube during drive.  I do think that the bias point would be very temperature sensitive with Frank’s circuit.  Perhaps his circuit with my bias circuit is a good mix there.

So I am considering having to do a major re-thinking on my circuit.  Right now, the feedback from the driver output is ac-coupled back.  I think it was Frank that may have suggested adding dc feedback and I also feel that this is the necessary way to proceed on my design.  To get everything needed, I think that my first schematic does basically represent the complexity needed to fulfill all requirements.

By the way, the FETs could probably be a smaller FET, like something in the IRF730 or IRF840 class.

All for now, Tom WA3KLR.
Logged

73 de Tom WA3KLR  AMI # 77   Amplitude Modulation - a force Now and for the Future!
nu2b
Guest
« Reply #58 on: November 03, 2005, 12:07:45 PM »

Tom-JJ,

Sorry, I forgot to address your question on the extra FET driving power.
The intent was just to indicate that the driver had to support (not supply)
more power than a regular push-pull grid driver.The FET can easily handle
having to sink the total cathode current (grid current,plate current
and pullup current).

Note: I'm still estimating higher voltages than the other guys since we
probably want to accomodate older tubes with lower emission.

The drive circuit power supply for the grid (approx +100 volts) only
supplies the grid driving power.

The pullup power supply (approx +300) only sources power for the
pullups (=20ma each =approx 1/2 the idling current value of 40ma each), and actually sinks

cathode current from the tube during cutoff.

The 3000V supplies only the plate current of course. It's interesting, however,
that with the FET in the cathode, biased up about 180-200 volts, the idling dissipation
of the tube is reduced slightly since its drop is now only 2800v. The remaining dissipation
is taken up by the FET. The full modulating swing of 3000v is still available
since during the peak current swing, the FET drain voltage is only 10-20 volts.

Regards,
BobbyT
Logged
nu2b
Guest
« Reply #59 on: November 03, 2005, 12:10:12 PM »

Frank,
To address your question on the need for pullups:

It would seem to be of great advantage if we had a modular design which could be
initially tested and aligned to the expected modulator requirements without actually
connecting to the tube or having to mess with 3KV in the vacinity.

The FET driver and power supply for the grid bias and the pullups can be connected
independently of the tube if the pullups are present. Additionally, we could connect
external test resistors (maybe in series with a diode and using a separate supply) in parallel with the pullups
to draw extra power to somewhat simulate the extra current drawn by the tube.
(This might be the way to check for temperature drift)

You could do the initial adjustment to obtain the required 70 volt differential
between Vg and Vk. This way, when connected to the tube, you are already in the
ballpark for the expected idling current. You can also run all of the initial
audio tests, add stray capacitance, etc. Doing this on the bench might be easier
than doing it in the rack, especially for an old fart like me.

Regards,
BobbyT
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #60 on: November 03, 2005, 12:17:23 PM »

Tom,
I agree on the FETs but the 900 volt units would make it a lot stronger.
I wonder what is the source of distortion? Would it be better to use lower resistor values in the phase splitted drain / source? Feed back should be easy to apply later.
Yes I noticed no input resistors in the op amp. Maybe the driver reisitor values are a bit low. Don't forget simulating the 833 with a pullup resistor will set the gain higher.
That could degrade distortion. The tube is going to be a variable resistor load.
Tom also look at using op amps driving high C loads. Usually you have to put another resistor in the gain loop in series with the op amp output with the feedback resistor outside the resistor. I have sen op amps take off without the resistor. Ive even seen oscillation with op amps driving shielded cable.
Logged
nu2b
Guest
« Reply #61 on: November 03, 2005, 12:42:41 PM »

A thought about bias stability:

From the tube curves, it appears that gm = 700ma/50v=14ma/volt

This implies that to achieve approx 50ma +/- 5 ma stability for
the idling current we require about +/- 0.3 volt stability on the bias.

This normally wouldn't be so bad, but since we are trying for approx 70 +/ 0.3volts
for Vgk, and Vgk now depends on both the FET drain-to-ground stability
and the +Vg supply stability it suggests that the +Vg supply might needs to be regulated
with perhaps a front panel small delta-V control available.

If not regulated, e.g use of basic line supply, maybe a very long time constant
differential tracking gizmo can be devised.

BobbyT-NU2B
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #62 on: November 03, 2005, 01:27:40 PM »

The whole driver circuit would run off a 7812 voltage regulator. Temperature compensation could be series diodes with 1 side of the bias pots. The RDS -on will increase with temperature so resting current will go down as heat goes up as long as the tube is stable. It could become a hum monster if the power isn't regulated.
Logged
2ZE
Guest
« Reply #63 on: November 03, 2005, 03:17:47 PM »

Frank, and All:

OK on all so far. What about a FET driver op amp, that is used to drive a High C input. I think it would be better than an op amp, but the FQA11N90's are pretty low C as fets are concerned.
I plan on using a small 10"x6" chassis and mounting it dirctly to the heatsink. I can mount a circuit on perfboard and mount it there. This will also house the transformer for the +300 vdc supply which I can regulate for any voltage 0-300 vdc. makes a nice neat package that I can mount to my mod. deck and keep the grid leads short.
Also Frank, the cathode reisitors I am using now for degen. feedbak are 12 ohms, so even better.
As soon as a design looks good, I can build. Might have to be next weekend now, but no biggie.

Mike
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #64 on: November 03, 2005, 04:14:01 PM »

Mike
I don't think you need 300 volts. You do need 12 volts or 24 volts regulated.
I see no reason for a high voltage pull up. A 50 K resistor across the fet will work as well as the resistor in a linear amplifier to hold a tube in cut off. So you have 12 ohms and I have 10 and 1 ohm. You could throw my circuit in with your present bias and test it low level. They you could slowly work your grid bias to zero and see what happens. Yes a fourth fet amplifier could be added for more voltage gain. 4 resistors and 1 cap and the fet. Same 12 or 24 volt reguulated supply. Take my circuit. Add 15 volt transorbs gate source each fet. Add more bypass caps as I said earlier.
Then add 50 K resistors across the FETs to limit voltage. Sounds like your rig would make a good test bed and the grid bias will keep it safe. Yea. just bolt the heat sink to the chassis and mount the parts inside. The little regulator will be the hardest part.  Feed back is no big deal just have to decide how much and where you want to connect it. We will run everything class A and see where it takes us. fc
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #65 on: November 03, 2005, 04:16:43 PM »

Mike you don't need no stinkin grid leads for this baby....grounded grid
The drive goes into the CT of each fil transformer. Then put a couple caps cT to each leg and some RF bypass to gnd.
Logged
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


"Let's go kayaking, Tommy!" - Yaz


« Reply #66 on: November 03, 2005, 05:59:15 PM »

We don't need no stinkin' grid bias anymore!   Grin

Hey Franz, I was wondering about the distortion figures Tom simulated. What is considered a good figure, under 1%?  I also will be curious what caused that. I notice that Tom's,  even with feedback, had over 2%, but maybe it was caused by the bias moving around.  Your bias was solid, but no neg feedback.

I wonder what a pair of 833A's ALONE generate for distortion figures?

Knowing that a -30db IMD linear with a good balanced modulator driver is around 0.1% makes me wonder.

It's really something to see three of the brightest engineers on this BB go at this project.  Thanks for the continuing efforts, guys. When we get a finished product, we'll see to it that  it gets well published so that anyone looking to upgrade their modulator can do so.  2ZEe and I will be initial fodder.

Mike, I'm picking up my 1KW mod tranny tonight at around 7PM. So might be back and on the air by 9PM. If not, then catch ya on tmw night for sure. 75M, of course.

T



Logged

Use an "AM Courtesy Filter" to limit transmit audio bandwidth  +-4.5 KHz, +-6.0 KHz or +-8.0 KHz when needed.  Easily done in DSP.

Wise Words : "I'm as old as I've ever been... and I'm as young as I'll ever be."

There's nothing like an old dog.
WA3VJB
Guest
« Reply #67 on: November 03, 2005, 09:04:32 PM »

Gary brought my attn to this thread. VERY sharp minds at play.
May I volunteer any aircheck recordings to post as MP3 for the group to hear?

Tom -- earlier in the thread you said --
Quote
Dave: Talk about internal tube shorts... I remember the time when Gary/INR had the IJY fet board driving the grids of his 833A modulators with a 4X1 final. G had sold the full rig to some African govt for a hefty price and was gonna ship it asap. Tron was visiting the shack and let out one last atomic yallo into the mike. A loud buzz rang out as an arc ripped it's way across the fet board, destroying every solid state device.  Grin Grin Grin

The historical record shows the rig in use at the time was the INR 4-400 rig.
The Pair 833A into 4X1 was the next job, and still soldiers on, tribanded at W2KBW.
Worked it and him recently on 40m.
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #68 on: November 04, 2005, 10:34:53 AM »

Tom,
I'm thinking about the distortion. I'm not sure where wer should put the feedback
The 1 and 10 ohm resistors put degenerative feedback on the tube. Heck Mike is running 12 ohms fixed. I would start with no feedback and see what happens.

Look at a grounded grid RF amp and consider what is happening here. Similar.
Brain working but pesky work is getting in the way today.  fc
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #69 on: November 04, 2005, 11:26:14 AM »

And I remember one Saturday afternoon when I got a call from a guy in Africa asking about the rig because it blew up. The IJY board had under sized heat sinks that caused the driver to blow up.
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #70 on: November 04, 2005, 11:27:53 AM »

Mike ZE can you measure your present distortion to get a real base line distortion number?
Logged
2ZE
Guest
« Reply #71 on: November 04, 2005, 05:50:42 PM »

its around 2-3%, I know.
Logged
WA1GFZ
Member

Online Online

Posts: 11152



« Reply #72 on: November 04, 2005, 09:47:55 PM »

So mike that is the base line. I'm thinking Tom klr simulation of pull up increased the gain a lot since the drain resistor was increased so high. This may have increased the distortion. one thing a pull up would do would be to put the stage more into class A
so who knows it may help on the turn off of the tube?Huh
Logged
Tom WA3KLR
Contributing
Member
*
Offline Offline

Posts: 2120



« Reply #73 on: November 04, 2005, 10:47:14 PM »

Guys,

I'll publish my new version schematic here tomorrow.  It is now a shunt dc voltage regulator configuration; same complexity as before.  The tube bias is regulated voltage, not current regulation.  You have the option of the same biass for both tubes or independent bias adjustment by the way you wire that circuit.

You need +15V, -5 V and +120 V.  The -5V is needed to get the common mode input range and output range required from the op amps.  The +120 V is for cathode pull-ups which I feel is needed.  Current on this supply is 24 mA. max.  Does not have to be regulated or super-pure dc, but stay above 120 V worst case.

FET dissipation (each) is 17 Watts worst case.  Only 2 FETs.

Q to Tom K1JJ - What audio level input do you need?
 Right now, the circuit requires about +12 dBm for max output.  This is probably a little high for most folks.

Sleep tight.
Logged

73 de Tom WA3KLR  AMI # 77   Amplitude Modulation - a force Now and for the Future!
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


"Let's go kayaking, Tommy!" - Yaz


« Reply #74 on: November 04, 2005, 11:47:13 PM »

OK, Tom -

+12dbm is FB for my set up here. I usually run it at 2 volts out.  +12dbm is 0.9 V, so plenty.  I have several adjustments anyway.

Mike is that OK wid you?

Tom, did you simulate your new circuit, and what did you see for distortion this time?

I'd be curious how you think Frank's circuit performance compares with your new version using more components and the two extra power supplies ?  I don't mind building one more complex if it reduces distortion, etc.

The separate bias controls is a good idea. Most of my tubes are pulls of questionable emission.  Never a matched set... Cheesy

T

Logged

Use an "AM Courtesy Filter" to limit transmit audio bandwidth  +-4.5 KHz, +-6.0 KHz or +-8.0 KHz when needed.  Easily done in DSP.

Wise Words : "I'm as old as I've ever been... and I'm as young as I'll ever be."

There's nothing like an old dog.
Pages: 1 2 [3] 4 ... 6   Go Up
  Print  
 
Jump to:  

AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands
 AMfone © 2001-2015
Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
Page created in 0.081 seconds with 18 queries.