803 tube for audio

[KE6DF]

Does anyone have information on using the 803 in AB2 or AB1 for audio?

I notice that there is nothing in the RCA datasheet (the only one I have).

Orr had a design for using a triode connected 803 as a modulator tube in the '59 Handbook -- similar to using an 813.

But it seems to me the 803 could be used grid driven in AB1 as well.

Anyone had experience using 803's as modulators?

[KM1H]

Tie the screen to the plate?? I dont think it would be veryy happy with 2-3KV

[KM1H]

Same problem

[KE6DF]

Actually, you don't tie the suppressors to the plate either.

The suppressors can only handle around 500 v max on 803's.

In the schematic you published, the suppressor is tied to the speech amp power supply at around 325V.

The circuit is similar to a triode connected 813 modulator, the difference being that the 803 needs some suppressor voltage to get enough plate current swing, whereas the 813 is  happy with the suppressor at 0v.

Thanks hro for posting the schematic. I think it's the one from the 1959 radio handbook I mentioned. In the article, they refer back to the speech amp circuit.

[KE6DF]

OK, we are probably shooting a dead horse in a barrel here, but the circuit you posted above says just what I said.

The suppressor connects over to the center tap of the driver transformer.

And the commentary on the circuit says the suppressor voltage is 280 - 340 vdc.

The Orr article suggests using a speech driver circuit with a pair of 6B4's -- and those would run on a voltage in this range.

[The Slab Bacon]

Attached below is the circuit for an 803 ultra mod final modulator.

803's will SCREAM! in this configuration. They use some slight cathode bias along with the control grid bias.

Take a close look at your skizmatic, methinks you gots an ooops in there!!   

[N8UH]

Attached below is the circuit for an 803 ultra mod final modulator.

803's will SCREAM! in this configuration. They use some slight cathode bias along with the control grid bias.

Take a close look at your skizmatic, methinks you gots an ooops in there!!   

Well, at least ONE of the 803's would scream... 

[WD8BIL]

WB6IYH said:

The suppressor connects over to the center tap of the driver transformer.

And the commentary on the circuit says the suppressor voltage is 280 - 340 vdc.

With the screen tied to the control grid and a positive voltage on the suppressor grid you've basically made a tetrode. The suppressor now becomes the screen!

[WD8BIL]

No, Brian.

He was referencing your original circuit posting.
I just pointed out the nature from another view point.
Helps understand things sometimes!

[W2PFY]

A Johnson Desk KW mod transformer works perfect

Never seen one of those for sale. I have a desk kw. what is the value of those resistors going to the grids?

[KE6DF]

Brian, thanks -- that looks like an nice circuit. I would need to find a mod transformer with taps for the screens, though.

WD8BIL -- that is an interesting way to look at the Orr circuit. Thanks for the comment.

Dave

[Bill, KD0HG]

OK, I've got the RCA Aug. 15, 1944 data sheets for the 803.

I prolly ought to scan them on request, but here's the nutshell:

803- RF Power amplifier Pentode-

Max ratings: DC Plate voltage- 2000 max volts
Grid 3- 500 max volts
Grid 2- 600 max volts
Plate current- 160 max mA.
Plate input- 180 max watts
Plate dissipation- 125 max watts.


----------------

TYPICAL OPERATION, Tetrode connection, grids 2 and 3 tied together:
Plate modulated Class C telephony
DC Plate voltage- 1600 volts
DC screen voltage- 130 volts (obtained through a 20 K ohm resistor to preferably unmodulated side of plate voltage supply)
DC plate current- 160 max mA
DC grid voltage- -180 volts (obtained from a grid resistor of 4000 ohms, or a combination of fixed grid bias and a resistor)
Peak RF grid voltage- 320 volts
DC plate current- 150 mA
DC screen current- 75 mA
DC grid current- 45 approximate mA
Driving power- 15 watts
Power output- 155 approximate watts

----------------------

There is no information on audio amp use of the 803, but there's the specs for single-ended Class B linear RF operation:

DC plate volts- 2000
DC screen volts- 600
DC suppressor grid volts- +40
DC grid bias volts- -40
Peak drive volts- ~50

Output power in Class B is not specified.

[WD5JKO]

I have never seen a P-P audio amp where the screen grid is powered via a high ohm dropping resistors that pulls power from a very high B+. This arrangement introduces some dynamics that need further review:

DC conditions: The zero signal screen voltage could get quite high depending on the bias set idle condition. With some tube types, the maximum screen voltage rating could easily be exceeded, and either socket flashover, or internal tube flashover could occur. From a DC standpoint, the screen voltage swing possible could be reduced by a resistor added to each screen to ground. This would form a voltage divider.

AC conditions: As audio drive is increased, the average screen voltage will drop due to a rise in screen current. This suggests the screen voltage at zero signal has to be high, and at maximum signal the average screen voltage needs to be high enough for the desired audio output in watts. A very high screen voltage at zero signal suggests that the grid 1 bias needs to be very negative.

The high screen grid impedance, if to just a high B+ will be a form of negative feedback. This scenario is described by Bacon, WA3WDR in May 1992 AMPX:

http://amfone.net/AMPX/100.htm

Here with a pair of 8417's he added 820 ohms to each screen and raised the screen voltage from 300V to 380V. He got more power at less distortion. Also had to set the bias higher.

Since this 803 modulator Brian has drawn up gets the screen voltage through a resistor going to a teritiary winding on the modulation transformer, we add a new twist to Bacon's circuit. The analysis of that is beyond me, but in lower voltage circuits with tubes like a 6L6, and no series resistor, this arrangement makes a tetrode act more like a triode (lower Rp) with the drive requirement between that of a triode and a tetrode.

I am not sure how the 803 modulator would work as shown, but I suspect the DC static conditions will be a challenge to hold in check. Adding a voltage divider as I stated above would help; but in doing so, the AC dynamics would be altered as well.

Jim
WD5JKO

[WD5JKO]

It is common practice with ULTRA LINEAR applications.

Brian,

The Ultra Linear concept does not use dropping resistors. Show me one example using a screen dropping resistor where the upstream screen voltage is way beyond (> 2X) the tube published maximum G2 voltage rating.

To do Ultra Linear with a tube like 803 or 813, you need an extra Center Tap winding on mod transformer with a lot less turns than the plate winding. Then return the CT to ~ 500-750V DC source. Think of two ART-13 mod transformers combined as an example (would need to be on the same core though).

Edit: Several Ultra Linear circuits by Acrosound:

http://www.dvq.com/hifi/images/acro.pdf

Notice how they do it with the 6146 amplifier...Just like I described above.


Jim
WD5JKO

[WD5JKO]

Brian,

  OK good luck with the cleaning, moving, and that first night. I hope that this is not like a boat or a swimming pool. By that I mean there are two glorious days, the day you get one, and the day they are gone. 

  Also cathode bias in audio power amps is usually limited to a class A amplifier, or maybe class AB1 when the average power is low, and the cathode resistor is heavily bypassed. The R-C time constant can hold the bias briefly for an occasional peak. Using cathode bias beyond ~ 30 watts with a tube amp is not practical.

  Case in point, I have an old Fisher amp with 7591's that uses cathode bias. Part of the cathode resistance is the 12AX7 low level tube filaments. I can crank the audio with music and see the filaments modulate to the music. Run it up with a sine wave, and the tubes each get up to 18v across pins 4.5.

Jim
WD5JKO

[KE6DF]

So I guess I'm wondering what are the real advantages of going to the ultra-linear circuit with the 803?

The first circuit brian posted (which is straight out of Orr's '59 handbook) seem simpler.

No taps needed on the mod transformer. No bias supply needed. There is a voltage on the suppressor, but this would be available from the speech amp as shown.

Seems and all around simpler approach.

Does the UL give better audio quality? Is that the reason?

I don't know about 803's, but plenty of people use triode connected 813's and get good audio from them.

[Bill, KD0HG]

So I guess I'm wondering what are the real advantages of going to the ultra-linear circuit with the 803?

The first circuit brian posted (which is straight out of Orr's '59 handbook) seem simpler.

No taps needed on the mod transformer. No bias supply needed. There is a voltage on the suppressor, but this would be available from the speech amp as shown.

Seems and all around simpler approach.

Does the UL give better audio quality? Is that the reason?

I don't know about 803's, but plenty of people use triode connected 813's and get good audio from them.

A modern UL approach where screen voltage is a lot less than plate voltage is to use high voltage zeners in series with the screen grids. There weren't such things back in the Williamson days. Then screen current becomes a small consideration.

I'd be hesitant to use any tube in a project where spares are tough to find. That might make 813s a better choice.

[WD5JKO]

A modern UL approach where screen voltage is a lot less than plate voltage is to use high voltage zeners in series with the screen grids. There weren't such things back in the Williamson days. Then screen current becomes a small consideration.

  This is quite interesting. So if the B+ were 2400v, and the desired screen voltage was 600v, a zener dropping string of ten 180 volt zeners in series would do, and be done for each tube. A couple of concerns pop into my head with this approach:

* A zener diode needs some minimum current, so a resistor at each tube screen to ground could be set for a minimum zener current of a milliamp or two when the tube is not drawing any screen current.

* If the tube draws 50ma or so maximum screen current, then the zener thermal issues could lead to high wattage, high voltage zener diodes. In the example above with ten 180V zeners, each would dump 9W of heat at 50ma. For voice work, maybe 5W zeners heatsinked to metal with thermal epoxy would do.

* Any AC ripple on the plate supply would also be on the screen supply at the same magnitude. This means AC hum on the modulation should the plate supply be anything but pure DC.

I am attaching a link and a file concerning UL operation of a tube P-P amplifier. The UL concept is great, and useful, but I feel it is not really applicable to the application that started this thread. For that, the Bill Orr circuits seems quite practical so long as one has a good number of 803's on hand.

http://www.aikenamps.com/UL.pdf

Jim
WD5JKO 

   

[KE6DF]

I'd be hesitant to use any tube in a project where spares are tough to find. That might make 813s a better choice.

I started collecting 803's a couple years ago because they seemed like an under appreciated (and therefore under priced) tube. So I have about a dozen.

They are not really all that rare. For example, as I type this there are 9 listed on epay.

The price does seem to be going up a bit, however, as mentioned earlier in this thread.

Dave

[Bill, KD0HG]

So Dave, what are your thoughts about your project at this point? We might be getting off track here but it is a good mental exercise.

Bill

[WD5JKO]

The link I provided from "Hafler and Karoes" had the following stipulation concerning Ultra Linear operation:

"It has been found that the screens must be fed from a low impedance source, or the benefits of this arrangement cannot be realized"

http://www.aikenamps.com/UL.pdf

  Looking at the circuits posted, the grid2 resistors were parasitic suppressors primarily, and likely 1K or less in resistance. They also add a little NFB like the Bacon circuit in the AMPX article I linked to. The one with  cathode resistors was a class A audio amp of about 30 watts depending on plate voltage used.

  How does one use cathode bias with a tube intended to be run Class B where the cathode current might idle at 50ma, and peak to 350ma? You just cannot get a stable operating point this way unless you go class A. Also, with the screen voltage wanting to come up approaching the plate voltage with no drive (no screen current), the bias need (at G1to cathode) will increase substantially. It is likely the bias would not be enough, and the tube would go into thermal runaway. That is why I originally brought up a voltage divider at the screens to limit the maximum G2 voltage to something safe. Using a teritiary winding on the mod transformer is better since "Hafler and Karoes" state that for UL operation, the screen supply impedance must be low, and 20K is surely anything but low.

  I hope that those reading do not perceive this discussion as a hi-jack of the original post. Although Brian, and I may not agree completely, we both think the Ultra Linear circuit is a wonderful thing. Now how we might go about it is where we have a difference in opinion. The circuits, and technical references brought up thus far should benefit all of us.

   I still feel that if we cannot do UL at high power simply, and reliably, then the original Bill Orr circuit is the way to go. Has anyone tried this circuit, and have any pros and cons concerning it? Same for UL at high power, anyone out there done it? Comments?


Jim
WD5JKO 

[Steve - WB3HUZ]

None for modulator use. Any distortion reduction is swamped out by mod tranny, general lack of modulation linearity in the final, and distortion in most people's receivers.

So I guess I'm wondering what are the real advantages of going to the ultra-linear circuit with the 803?

The first circuit brian posted (which is straight out of Orr's '59 handbook) seem simpler.

No taps needed on the mod transformer. No bias supply needed. There is a voltage on the suppressor, but this would be available from the speech amp as shown.

Seems and all around simpler approach.

Does the UL give better audio quality? Is that the reason?

I don't know about 803's, but plenty of people use triode connected 813's and get good audio from them.

[Opcom]

For high voltage Zeners, a high wattage NPN power transistor can assist well. Because the Hfe of the transistor may act to further reduce Zener current, the aforementioned ~1mA resistor can be connected between the the transistor base and emitter. Using a transistor to assist the Zener also opens the opportunity to decouple hum with a capacitor at the base. Its filtering value would be multiplied by the Hfe of the transistor.



I disagree with the claim that it is not important to remove distortion for modulation use. Issues of the mod tranny and general lack of modulation linearity in the final can be reduced. Everyone uses voltage feedback and ignores current (therefore ignoring power), assuming it will follow the voltage but as just pointed out, it does not always. If the RF stage has modulation nonlinearities, can they be reduced?

The first step is to have as clean and well-regulated a modulator as possible. It may mean a UL circuit or an oversized mod. with a light load.

The next step is to get feedback from the modulated stage (with a detector) and apply it. If the modulator uses plenty of cathode feedback or another brute force UL scheme where just the transformer and the modulator tubes are in that loop, the phase shift across the modulation transformer should not be so great that it would cause instability. An objection might be that the RF stage output average level could change by 3dB depending on loading and drive and this would increase or reduce the 'dB' of the negative feedback being used to pre-distort the modulator output. That is true but the RF envelope should and would still be forced to follow the audio envelope fed to the modulator.

Modulation transformers are not usually would with an extra winding for UL schemes but other ways of using power feedback (not just voltage) might suggest themselves. It has to have been done by smart people before.

Nothing can be done about people's receivers, but reducing distortion to at least the level of an old BC rig (or less) is not unreasonable. If someone is building and not buying, it seems more along the lines of good practice to spend a little more time and try to have the cleanest and most accurately reproduced transmission possible, no matter what scheme is used.

The next argument might be that with all the speech processing and clipping, the voice signal is distorted anyway. That is true, but that distortion is intentional and controlled. Why permit uncontrolled distortion because of an unruly modulator and RF stage?

It is not strictly necessary to remove as much distortion as possible, but the attempt and work to do so should not be discouraged. We don't live anymore in the time of those old 803's but we can apply new ideas to the fine old tubes and circuits and make them better.

[Steve - WB3HUZ]

Consider that no BC rigs ever used ultra-linear modulator topology. Given that some BC TX manufacturers were more than given to over-kill, this should tell you something very important about the value of UL.

[W3SLK]

Steve said:

Consider that no BC rigs ever used ultra-linear modulator topology. Given that some BC TX manufacturers were more than given to over-kill, this should tell you something very important about the value of UL.

Pretty much "game, set, & match!"