The AM Forum
March 29, 2024, 03:23:20 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 ... 6   Go Down
  Print  
Author Topic: Simple Audio Driver for 833A modulator  (Read 94340 times)
0 Members and 1 Guest are viewing this topic.
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


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


« Reply #25 on: November 01, 2005, 01:30:51 PM »

what kind of file do you need i can convert

Bill/IFR said he can post it from his home website. I just sent it to him. We should be in business soon.  Thanks, Bill.

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

Offline Offline

Posts: 11152



« Reply #26 on: November 01, 2005, 02:24:26 PM »

Tom A good transformer should have good common mode rejection. The cap valuse would depend on how much you have. Also the CT balance would effect leakage. I suspect you can ground the transformer and if you are lucky it has a faraday shield. I didn't include any frequency roll off that can be added later.
As BobbyT said a pair of shunt regulators driven by a phase splitter. Two pots set the regulation point Keep the temperature stable and it will be fine. K.I.S.S.
Logged
kc2ifr
Guest
« Reply #27 on: November 01, 2005, 02:49:35 PM »

Hi Tom,
Did not receive it as yet. Its 2:48PM....what time did u send it???
Just in case...my email addy
wgr@adelphia.net
Logged
WA1GFZ
Member

Offline Offline

Posts: 11152



« Reply #28 on: November 01, 2005, 03:50:30 PM »

Bill,
I just sent jpeg and bmp to you. fc
Logged
WA1GFZ
Member

Offline Offline

Posts: 11152



« Reply #29 on: November 01, 2005, 03:54:52 PM »

I see Bobbyt has a bypass to ground at the cathode. Good idea to keep the tube from oscillating.
Logged
kc2ifr
Guest
« Reply #30 on: November 01, 2005, 03:59:58 PM »

Here is the pic.....I changed it to a jpg file...
www.kc2ifr.com/photogallery/833a_driverbmp.jpg
Logged
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


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


« Reply #31 on: November 01, 2005, 04:15:18 PM »

Thanks again, Bill. Came out FB.  I originaly sent the .bmp back to you a few minutes after your offer. O'well, lost in cyberspace.

Frank: Reply to your comments:

Tom A good transformer should have good common mode rejection. The cap valuse would depend on how much you have.

*****  How much I have of what... capacitance value?  Name a value. ***

Also the CT balance would effect leakage. I suspect you can ground the transformer and if you are lucky it has a faraday shield.

**** OK, so you're saying to mount the fil transformers directly to the chassis. ***


I didn't include any frequency roll off that can be added later.

*** Freq roll off for what reason... stuff above 10kc generated by the fet driver stages?  ***


As BobbyT said a pair of shunt regulators driven by a phase splitter. Two pots set the regulation point Keep the temperature stable and it will be fine. K.I.S.S.

** I've gotten a number of emails from guys wanting to try this. I'm thinking it would have been a common technique back in the 50's *IF* they had Fets then. For some reason, it didn't catch on once high voltage MosFets were developed..***

*** BTW, I plan to get on with Mike/ZE later tonight after 7PM on 75M, if the band is good, for an 833A discussion. You might join us if you're around Frank. Mike is keenly interested in doing this driver for his 833A rig 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

Offline Offline

Posts: 11152



« Reply #32 on: November 01, 2005, 04:20:50 PM »

It would be a good idea for someone to simulate it. switchcad should work.
This way there are no surprises. Roll off would be easy cap gate source.
As you know these fets work well into HF so some roll off might be a good idea.
Caps across the fil transformer .22 for starters. My guess. Radio shack sells diffent film caps.  fc
Logged
Bacon, WA3WDR
Member

Offline Offline

Posts: 881



« Reply #33 on: November 01, 2005, 09:32:21 PM »

I'm liking it... yes, I think you're right, the 833A in AB1 is a good way to go, and it looks like it doesn't even need a driver supply.  I hope the cathode capacitance (rf bypass and filament transformer capacitance to ground) doesn't kill the top end.  Precompensation and negative feedback could probably fix that, if it's even an issue.

This basic approach should be possible at lower power levels with 812s, tetrodes, etc.  Hybrid classic.
Logged

Truth can be stranger than fiction.  But fiction can be pretty strange, too!
WA1GFZ
Member

Offline Offline

Posts: 11152



« Reply #34 on: November 02, 2005, 08:10:11 AM »

The fil bypass would go between each side and CT. An RF bypass of low value to ground. The only C would be the transformer but this is a low Z circuit so who knows if it will have an effect. I'm thinking lower resistor values would be better at each gate to keep it RF proof. Noninductive resistors for the 1 and 10 ohm so as not to set up a tank.
someone should simulate it.
Logged
nu2b
Guest
« Reply #35 on: November 02, 2005, 08:54:23 AM »

Hey Guys... Just as a reality check note.

Ballpark 833A 500W design from a quick look at the tube curves:

Po = Ipeak*(Ep-Vpmin)/2 = 0.7*(3000-300)/2 = 945 W
This assumes some mod tranny loss and a bit extra for pos peaks or compression.

Eff = 78.5(1-Vpmin/Ep)=78.5*.9 = 70.5%
RL = 4*(Ep-Vpmin)/Ipeak = 4*2700/0.7 = 15.4k
Popeak = Ipeak*(Ep-Vpmin) = 0.7*2700 = 1890 W
Pavg (check) = Popeak/2 = 945 W
DC supply = Po/Eff = 945/0.705 = 1350 W = 3000V at 0.45A
Pdiss = 1350 -945 = 405W = 203W per tube

Pdrv-peak = Vgpk*Igpk =125*0.2 =25W
Pdrv-avg =Pdrv-pk/2 =12.5W

The above drive power is for classic grid drive not for cathode drive.
Our discussion here is more like a grounded grid Class B Linyear.

As such, we have to support both grid and cathode current.
therefore Ipk grid + cathode = 0.2 + 0.7 = 0.9A.
Looks like the mod driver has to support approx 60Wavg and 120Wpeak.
of which 12.5W is for the grid and the rest is added to the Pout.

Sounds like a 15W solid state block with reversed audio xfmr
might really be the way to go after all.

(You can do the basic audio block with a 40 year old design using
a 25.2v fil xfmr dc supply and about 5 real cheap xstrs plus audio xfmr)

Of course, I may also just be having a senior moment.
Isn't this fun or what!

Regards,
BobbyT   
Logged
WA1GFZ
Member

Offline Offline

Posts: 11152



« Reply #36 on: November 02, 2005, 09:16:40 AM »

Tom Vu wants no transformers and grounded grid so that was where we started. Still good idea to keep the other method in brain.
Logged
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


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


« Reply #37 on: November 02, 2005, 09:55:04 AM »

Hey Guys... Just as a reality check note.
As such, we have to support both grid and cathode current.
therefore Ipk grid + cathode = 0.2 + 0.7 = 0.9A.
Looks like the mod driver has to support approx 60Wavg and 120Wpeak.
of which 12.5W is for the grid and the rest is added to the Pout.
BobbyT   

Thanks for the schematic you sent email, Bobby!

I have a question about the data above.... since we're using 900V, 10A fets in the cathode, why is it a concern that it is switching 120W? Are you talking about power efficiency of 15w to the grid vs: 120 watt losses associated with the analog fet in the cathode or something else?   I suppose if things get warm we can use a bigger heat sink or simply parallel another fet.  Though, the lack of a bias supply with bias voltages on the fets, and inherent negative feedback of cathode drive with just one 12V supply attracts me more.

BTW, I received another great schematic for some improvements upon Frank's original design from Tom, WA3KLR. It uses low distortion op amps as Fet drivers, negative feedback to these drivers and a regulator chip/ fet to prevent 833A bias drift. This has great potential, I think. A little more complex, but appears well worth it. He said it probably needs tweaking.

I am posting here as a pdf file. Click on the the WA3KLR .pdf file below....  Check it out and please critique. Tom desires a simulation of it. Let's improve upon it.

73,
Tom, K1JJ

* WA3KLR_FET833dvr1.pdf (13.4 KB - downloaded 555 times.)
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.
2ZE
Guest
« Reply #38 on: November 02, 2005, 10:23:27 AM »

I like Tom 'KLR's design alot. I like the bias regulation. What OP amp does he use?
TLO74 or maybe a 5532 would work.
Frank's is so simple however.
As soon as someone sim's it, I have the parts ready to go (from my defunct E-PDM project). Could have it on as early as this weekend.

Gotta love the the brain power on this board!

Mike
W2ZE
Logged
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


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


« Reply #39 on: November 02, 2005, 10:33:41 AM »

Hola Mickey -
What a great time on the air last night, huh?  True ham radio, exchanging "vital" information. I was disappointed that no one even laughed at my Limbo song. All shop talk... Grin  We gotta do it again, soon.

Anyway, Tom speced using Dual TL 072BCP precision op amps. I think that means ALL the chips including the ones labeled "Fet driver"?

The other three big fets can be the standard 10A, 900 V variety for $250 each. [two cathode drivers and regulator]

He mentioned that he thought there wud be distortion WITHOUT the bias regulation, so added it. Can't hurt.  I'm pretty sure this can be pulled off without any bias on the 833A grids, just grounded. Imagine, just one 12V supply and the whole driver works.  How about that neg feedback loop around the op amps?  I think that with the cathode inherent feedback and this loop, that's all we will need for a super clean total modulator system.

Now we need someone to simulate it.

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.
2ZE
Guest
« Reply #40 on: November 02, 2005, 10:48:28 AM »

I guess I'm being a dunce here ( woudn't be the first time). How are the tubes getting biased? Are they being biased from the cathode? and how would it be adjustable?
I would like to have control over the idling current, and I just am not seeing it.

Also, how big do the 1 ohm resistors need to be?
also standard gate, drian and source busses need to be used( heavy copper connections)

I actually need to place a DIGI-KEY order for work today, and I will just piggy back my parts onto it.

the only thing different I am going to do is have another OP-amp stage right at the audio input so it is balanced in.

Please advise??

PS - I think everyone was a little "poker faced" with tech talk last night. Thats what happens sometimes. However, it was better to have everyone on the same page discussing one common topic, lowers the risk of QSO implosion ten fold.
Logged
WA1GFZ
Member

Offline Offline

Posts: 11152



« Reply #41 on: November 02, 2005, 10:56:37 AM »

Tom Vu,
Resting cathode bias is set with current flowing through FET to provide a bit of positive resting voltage on the cathode. Remember the tube would be saturated if the fet was full on....like my pdm driver. The fet needs to be in the linear mode so it has monkey swingage. When the fet is in linear mode it is acting like a variable resistor.
It is also making some heat. Say you need to bias at 50 volts at .5A then the fet is making 25 watts of heat at resting carrier. It swings between full on and full off.
Resting carrier is probably worse case dissipation.  fc
Logged
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


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


« Reply #42 on: November 02, 2005, 11:03:15 AM »

I guess I'm being a dunce here ( woudn't be the first time). How are the tubes getting biased? Are they being biased from the cathode? and how would it be adjustable?
I would like to have control over the idling current, and I just am not seeing it.

Please advise??

Please advise, wilco, over and out...
So solly glashoppa, there IS no bias adjustment. You run them balls to the walls in class AAA until they puke.   Grin Grin

Actually, the tubes are DC biased with cathode bias, controlled as described below.  The dynamic audio voltage [fet] swings from ~ +70V to zero and this equates to zero to -70V referenced to the grid.  -70V = cutoff and 0V = 900ma or so. Mucho monkey swing. All depends upon input audio level, of course.

Mike, your'e looking at the new schematic, right? Take a look at the 78L05 regulator chip at the far right. R5 should really be an adjustable pot. This controls Fet M3 and IC  U1 that controls the gate DC bias of each cathode driver fet. C1 and C3 accepts the audio voltage riding on top.

Needs simulation before you build it. I so vely vely glad you be the first man on Earth to build it on a working set of 833A's. Get the bugs out for Mr. Vu.

Tom/KLR just sent me an email saying he intends to make more improvements on it and make more comments. No one is to build it yet. He will try to tailor it more to the 833A. All  three fets go on the same heatsink. More later. Stay tuned.

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

Offline Offline

Posts: 11152



« Reply #43 on: November 02, 2005, 11:44:28 AM »

Tom KLR
Change feedback to a resistive divider. The 833 should be off at 100 volts on the
cathode. The input of the op amp needs to be 5 or so volts at carrier. A zener clamp or diode clamp could protect the op amp from over voltage at input. This eliminates the caps and 2 resistors in the input bias circuit. This does add the 100 Volt pull up voltage. My circuit I did the series the series resistor to fix the gain at Rd/Rs. I wondered if this would introduce distortion but could not see how. I have seen it done cascaded fet amplifiers. The upper FET has a grounded gate. We all know a FET and triode are the same thing. KLR circuit controlls the drain voltage with an active loop so more complicated but sexy.
I had a boss once who thought a design with the least number of parts was the best.
Lets see how far we can go with it.
Logged
2ZE
Guest
« Reply #44 on: November 02, 2005, 11:51:19 AM »

Quote
Actually, the tubes are DC biased with cathode bias, controlled as described below.  The dynamic audio voltage [fet] swings from ~ +70V to zero and this equates to zero to -70V referenced to the grid.  -70V = cutoff and 0V = 900ma or so. Mucho monkey swing. All depends upon input audio level, of course.

AHHH HAA!
I semen.
Didn't study schematic close enough to see that Tom mentions adjustment, but didn't draw the actual ass-wiper on the pot. Sooo Solly mastel KLR.

Getting parts together, and will standby for further updates.

Like a fat kid playing dodgeball....
I'm Out!
Logged
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


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


« Reply #45 on: November 02, 2005, 03:05:23 PM »

I was trying to see the difference between the two circuits .. why Tom/KLR wanted a
regulated  bias adj. Now I see that it's for thermal drift of the cathode Fets. The
regulator fet is on the same heatsink and compensates.  So, has nothing to
do with dynamic distortion.

So, if the fets are thermally stable enuff, and I suspect with a range of
0-70V being so huge and loose, and a big heatsink, we might not need the thermal compensation.

The feedback Frank said he cud add with a resistive divider. The remaining question is if precision op amps will make any difference at all in performance. I suspect you'd never notice the diff, but that's open for debate.

These observations are not fact - I'm just trying to generate discussion and maybe pull the best features into one circuit. Frank's is certainly simpler in parts. We'll have to see in simulation what they do. Frank mentioned he has loaded it into Switch-cad for a simulation later.  We're making progress.

T


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

Offline Offline

Posts: 11152



« Reply #46 on: November 02, 2005, 03:50:59 PM »

I'm stumped how do I simulate an 833 hanging off the drains.
any suggestions? No triodes in th epackage.
Somewhere in th ehouse I have a circuit of a fet driving a tube. I need to put on the miner's helmet.
Logged
nu2b
Guest
« Reply #47 on: November 02, 2005, 05:15:29 PM »

Frank,
In actuality, large tubes are almost direct physical scaleups of small tubes.
Relative physical spacings-etc.

Some times the low power models in spice are not power limited.

I've used a 6L6 at high voltage to simulate a 4-400 for example.
I just tried the 12AX7 and 12AU7 models and generated a kilowatt
of audio per my previous ballpark design thread.
The bias is not quite right, but close enough for having fun.
I normally use the Circuitmaker program.

Regards,
BobbyT

PS-i did find the old schematic for the 30watt 6L6 amp using FET's in the cathode.
I'll try posting it here as a .wmf file. No guarantees-it's just for fun

Note:The .wmf file didn't attach
Logged
nu2b
Guest
« Reply #48 on: November 02, 2005, 05:45:06 PM »

Just another question:
Remember that the opposite plate is swinging to at least 6000V.
doesn't that mean you'll need at least 150 to 200v to ensure cutoff.

Maybe bias at +100V with FET pullups to +300.

The pullups probably will be needed, since the tube will appear as a lower value current source near cutoff thus with high frequency 10KC audio and a bit of filament capacity to ground may cause loss of the cutoff peak value and cause the tube extra dissapation.

Regards,
BobbyT


Logged
WA1GFZ
Member

Offline Offline

Posts: 11152



« Reply #49 on: November 02, 2005, 09:27:23 PM »

Good piont,
The transformer will swing the voltage to double so yes more positive bias is required.
Pullup: consider this my GG triode connected 4cx3000A linear plate current dropped to zero with only a 50K series resistor with no pull up required (6KV). The FET near cut off will be a very high resistance so a low current will produce plenty of bias.
maybe the fet can be shunted by a 50 K resistor to limit voltage swing. I also put a fuse in the cathode circuit.
I still can't convince myself that the pullup is required. I've thought up a different configuration part mine part Tom KLR. A op amp driving a phase splitter to the two driver FETs with feedback around the whole modulator. Still this adds complication.
mulit op amps in one package has been a problem with high level signals. We actually measured cross talk across a quad op amp die at work. Head spinning after an hour and a half helping my daughter with Algebra homework. Word problems are a trip.
Logged
Pages: 1 [2] 3 ... 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 0.078 seconds with 19 queries.