Pages: [1] 2 3   Go Down

Author Topic: 40M QRP amp from 807s?  (Read 1149 times)

AC9QC

  • Posts: 57
    • HomeURL
40M QRP amp from 807s?
« on: September 16, 2021, 06:45:15 PM »

I'm wanting to build an amp for my OHR-100A using a pair of the 807s I have laying around (I have probably 10ish at this point, may as well use a couple.) I have a few transformers that should be able to do what I need from a power perspective, I probably need to get some tuning bits but I do have a roller inductor flopping around somewhere too.

The data sheets I've seen use a combination of grid and cathode bias, but I'm planning on trying fixed bias. Any reason not to?

A big question I had was on the output circuit. I was deciding to see how they looked typically by checking the schematics from the HW-32A and the HW-101 to just get a generalized layout of the land. They run the tubes in parallel not push/pull. Is there a reason for that? Just trying to figure out what the benefit of it is to do parallel single ended vs P/P. Still trying to sort things out, I think as a mono bander this should be one of the simplest amps I could build. Looking at the RCA datasheet I'm expecting anywhere from 50-80W depending on the transformer (hence the anode voltage) I run. I may run them at ICAS too since I have such a big stash and would likely never use them otherwise if I don't have a power transformer of a lower voltage.

I'd also be curious on opinions of screen regulation, I had made a power supply at one point that was loosely based on an HP meter's which was a shunt regulator using a tube or two. Am I overkilling on the PSU for a CW amp? I may try this on a ubitx later so I don't mind if it's not overkill for that.

If I did decide to run SSB with this project, I figure it would likely be a rebias or maybe voltage changes at worse for the band I make it on, perhaps some plug in coils away from multi band operation. I was wondering what opinions were on linearity with 807s. It looks as if the datasheets I have run it in class C for telephony. I'm presuming that hurts linearity and I'd want to try to run it in class AB? Not trying to squeeze every last watt out but figured that I could probably get some decent gains on a ubitx this way.

This isn't a super serious project, the idea is to take my time and build something and enjoy the process.
« Last Edit: September 16, 2021, 06:50:20 PM by AC9QC »
Logged

G3RZP

  • Member
  • Posts: 2254
Re: 40M QRP amp from 807s?
« Reply #1 on: September 17, 2021, 04:04:54 AM »

Single ended with tubes in parallel is easier, especially when using a pi tank circuit.  You don't need a split stator tuning capacitor, or a tapped coil, and with a pi tank, one high voltage variable and a low voltage bigger variable.

The voltages for AF amplification in AB1 or AB2 can be used: AB1 is more linear for SSB. Depending what make they are, you will probably have to over-run them very heavily to destroy them. 807s, depending on make, are very robust in my experience. AB2 needs  a 'solid' bias supply because of drawing grid current. The screen supply can be stabilised with a couple of VR tubes for SSB: I've never seen 807s sourcing  negative screen current, while for CW, with fixed bias, a potential divider from the plate supply is adequate. I prefer fixed bias: for RF, cathode bias doesn't do a lot for you. I would suggest that for SSB use, don't go above 1kV on the plate

They are a very versatile tube: they were even used in VFOs, as in the RAF type 53 aircraft tx and the Wilcox-Gay vfo for the RCA ET4336 tx. The crystal oscillator for that used an 807, too. The just post WW2 Marconi 350 watt input TGS541 used six 807s in push - pull parallel in the PA, driven by two in parallel, modulated by four in push parallel, and with another 3 in the 'absorber' keying (this was a circuit which in 'key up' had resistors across the HV line - 450 volts in that case - to stop the HV line drifting up in voltage.) The power supply used twelve (!) 5U4G in a bridge rectifier.  That tx for 350 watts in stood 6.5 feet high, 2 feet 3 inches wide and 3 feet front to back and used a total of five different valves, weighing in at 450 pounds......The Marconi Instruments TF867 signal generator used one in its oscillator (a bit OTT for 20 milliwatts of RF!) and the Cossor 339 oscilloscope used 807's in its Y amplifier. During WW2, intercept stations used 807 cathode followers for splitting feeds to several receivers: they were designed by 'Dud' Charman, G6CJ. Some early UK TV sets in the 1940s used them as horizontal sweep tubes.....and a number were used in stabilised power supplies.

Good luck with the project.

73

Peter G3RZP

Logged

AC9QC

  • Posts: 57
    • HomeURL
Re: 40M QRP amp from 807s?
« Reply #2 on: September 17, 2021, 11:12:02 AM »

Thanks for the reply, I'm working through some thoughts so far. It sounds like I won't be without some juice with what I have since most of them were RCAs from memory I don't think they'd be the cheesy ones which I suspect would be later, off shore variants or similar. I'd hate to think of the crystal current on an 807 oscillator!

It's also good to know that the audio data will get me started should I wind up looking at doing some SSB.

My plans for a bias supply are likely going to be something along the lines of a 24V doorbell transformer with a doubler or tripler circuit then perhaps followed by some form of linear regulator and a shunt.

If you don't mind some further questions, any advice about tank circuit Q? I'm thinking about 10T of #12 wire and something like a 10-100 ohm variable inductor which should get me to 7mhz without too much issue. I may have something smaller but I think that's what the junk box is going to put out for this venture. The 3.6 uH gives me something like 158 ohms of reactance at 7mhz and based on the wire resistance that'd give me a bandwidth of about 260hz. Is that going to be tight enough for CW, or do I need to tighten that up some more?
« Last Edit: September 17, 2021, 11:14:46 AM by AC9QC »
Logged

G3RZP

  • Member
  • Posts: 2254
Re: 40M QRP amp from 807s?
« Reply #3 on: September 17, 2021, 04:46:16 PM »

 First of all,

Working Q of a pi tank is determined by the component values, and you usually figure on a working Q  value between 10 and 20. (It is NOT critical!) You need to look up the formula in the various handbooks for values - if you REALLY have problems, i can provide the formulae.  The ratio of the Q when loaded to the Q when unloaded - i.e. no resistances - determines tank circuit efficiency. The higher the working Q, the greater the circulating current and since resistive losses are I squared R, doubling the circuit Q pushes the losses up by a factor of 4. Which is why the compromise of 10 to 20, and I believe that W8JI would live with a bigger range. So at HF, tank Q is not affecting close in signal bandwidth.

You don't say what Class (AB1 linear or Class C ) you are looking to work in, or what voltages you are proposing to use. You should be able to get away without a variable inductor, values in tank circuits are NOT that critical!! A danger with variable inductors is parasitic resonances when using the ends of the coil...

BTW, the frequency is 7MHz, not 7mhz ....mHz is millihertz -, one thousandth of a Hertz.....The difference may be obvious here, but it isn't  always!
Logged

KM1H

  • Member
  • Posts: 11155
Re: 40M QRP amp from 807s?
« Reply #4 on: September 17, 2021, 06:22:57 PM »

Quote
BTW, the frequency is 7MHz, not 7mhz ....mHz is millihertz -, one thousandth of a Hertz.....The difference may be obvious here, but it isn't  always!

In before our Correctness Troll, congrats.
Logged

WB6BYU

  • Member
  • Posts: 20896
    • Practical Antennas
Re: 40M QRP amp from 807s?
« Reply #5 on: September 17, 2021, 07:28:44 PM »

Quote from: AC9QC

...It looks as if the datasheets I have run it in class C for telephony. I'm presuming that hurts linearity and I'd want to try to run it in class AB? ..




Class C was for plate-modulated AM phone, where the tube
was only amplifying the carrier, not the modulated signal.

For SSB you certainly need AB1 or AB2 instead, and the
original data sheets probably list that only for audio (and
possibly for two tubes in push-pull).  If you are thinking
you might run SSB, then I'd recommend designing it
for AB1, then add an option to run it AB2 or C for CW
if you want to.

Note that the plate load impedance may change if you
are trying to get more output power for CW.

IIRC the 807 doesn't work well in grounded grid (unless
you modify the grid connection) but should work well as
a grid-driven stage, possibly with resistive loading on the
grid for stability.  My old Handbook may have had a couple
circuits using the 807, before they got replaced by the
newer 6146.  But any of the 6146 or similar stages should
give you a good idea of the circuitry, even if the bias
levels are a bit different and the output less with the 807.

The HW-12 / 22 / 32 SSB monobanders had an output tuning
capacitor with a fixed load capacitor, assuming the load was
close enough to 50 ohms.  That saves a big variable cap.  The
tuning cap in my HW-12 was fairly small, as I remember it,
so I assume that there was a lot of fixed capacitance in parallel,
and the tuning range limited accordingly.

G3RZP

  • Member
  • Posts: 2254
Re: 40M QRP amp from 807s?
« Reply #6 on: September 18, 2021, 04:08:08 AM »

WB6BYU Said:


Quote
Class C was for plate-modulated AM phone, where the tube
was only amplifying the carrier, not the modulated signal.

This is true, but my father told me about an experimental RAF approach that was never successful and never went into service using an 8018 (selected 807 for full output at 120 MHz) in a mixture of Class B and C as a 'linear' PA. Since even to this day, aircraft VHF always seems to me to be 9 parts distortion, maybe it didn't make  much difference! The British Army had a four 807 amplifier to follow their  #19 set: the 807s had cathode bias resistors, ran very hot, dragged  lots of amps out of the 12volt DC supply batteries and provided maybe 40 to 50 watts PEP....

A lot of old tube data sheets quote "Class B ratings" which are for AM linear service...

A snippet in a QST for about 1954 reported on a G station using an 807 in zero bias grounded grid - control and screen grids grounded. As it was only one, and only for 80m, he probably got away with it. However, for zero bias in AF applications, g1 was fed from g2 via an 18kohm resistor and drive applied to g2 - presumably the resistor was to limit the current in g1.
Logged

KM1H

  • Member
  • Posts: 11155
Re: 40M QRP amp from 807s?
« Reply #7 on: September 18, 2021, 01:12:44 PM »

Quote
The HW-12 / 22 / 32 SSB monobanders had an output tuning
capacitor with a fixed load capacitor, assuming the load was
close enough to 50 ohms.  That saves a big variable cap.  The
tuning cap in my HW-12 was fairly small, as I remember it,
so I assume that there was a lot of fixed capacitance in parallel,
and the tuning range limited accordingly.

The Heath HA-14 KW mobile amp did it with just a single 350pf ceramic doorknob. They must have figured that a pair of 572B's could tolerate any mismatch.

Carl
Logged

N2EY

  • Member
  • Posts: 5698
Re: 40M QRP amp from 807s?
« Reply #8 on: September 19, 2021, 05:48:08 AM »

The Heathkit HW-12/22/32 monobanders used fixed loading and needed antenna systems quite close to 50 ohms resistive. Since they only covered part of one band each, the pi-network could be optimized for those frequencies. Note that when those rigs were in production, 75 meters was only 200 kHz wide and 40 meter 'phone was only 100 kHz wide in the USA.

The Heathkit HW-16 also used fixed loading and needed antenna systems quite close to 50 ohms resistive. It only covered the first 250 kHz of 80, 40 and 15 meters, and its pi network was optimized for the three bands by switching in various fixed capacitors.

The Heathkit HA-14 "Kompact Kilowatt" also used fixed loading and needed antenna systems quite close to 50 ohms resistive. Unlike the above rigs, the loading capacitor is a single 350 pF fixed capacitor on all bands except 80/75, where an additional 500 of fixed capacitor is switched in.

Heathkit did this "shortcut" in the HA-14 to save money and space; that amp only cost $100 (without power supply) when it came out. It was primarily meant for mobile use, which is why the power supply is separate and the unit is so small. There are other "shortcuts" in the HA-14; it's not a good design to emulate.

IMHO, the best use of the HA-14 today is conversion to 6 meters. Or, if several can be obtained, convert to single-band use for quick band change without retuning.

This design "shortcut" results in the loaded Q of the pi network being very different on different bands. The loaded Q is lowest on 40 meters and highest on 10 meters.

Some might ask why all the focus on the loaded Q of a pi network. Here's why it's important:

1) the higher the loaded Q, the greater the harmonic rejection. This is good, but it has its downside:

2) the higher the loaded Q, the greater the circulating (reactive) current when using real-world components. This means greater losses and a need for heavier-duty components like switches.

3) the higher the loaded Q, the more you have to readjust the output network when changing frequency.

The Q of the output pi network does NOT affect the "match" to the final tubes as long as the network is properly adjusted.

Most hollow-state amateur transmitters use pi networks with loaded Q in the range of 10 to 20. It is left as an exercise for the reader to compute the loaded Q of the HA-14 on different bands.

73 de Jim, N2EY


Logged

AC9QC

  • Posts: 57
    • HomeURL
Re: 40M QRP amp from 807s?
« Reply #9 on: September 20, 2021, 11:48:37 AM »

Thanks all for the replies so far! Think I'm getting somewhere-ish. With my books in storage my research capacity is somewhat limited. I've got a bunch of ARRL handbooks and Ohr Handbooks but they're all away right now. Trying to search for push pull amps is somewhat of a headache due to all the audio stuff clogging the search engines. I think this will probably give me a rough place to go with the output circuit.

https://qsl.net/ve7sl/jones%20oscillator.html

This is probably the rendition I'd start with because it doesn't use a tapped capacitor which simplifies my parts roster some. I will say I'm tempted to buy a 2X100 variable off Ebay, there are a few nice ones there. It would likely not be in the spirit of the nature of the project though.



Obviously I wouldn't really use the input side, I kind of cringe looking at that just because I'd think it would flat out hammer a crystal. The power supply and bias would obviously look a lot different as well.

The perspective on AM is fascinating btw. I never really perceived that in a plate modulated setup the RF side was in reality just an oscillator from a linearity standpoint. It seems so obvious in retrospect.
Logged

KB2WIG

  • Member
  • Posts: 777
Re: 40M QRP amp from 807s?
« Reply #10 on: September 23, 2021, 06:30:43 AM »



If books being 'away', the following may be of comfort,

http://www.tubebooks.org/technical_books_online.htm

A few of the Orr West Coast Handbook are there.

This probable should be a 'stickie' in the Boatanchor section....

KLC
Logged
EXTRALight  1/3 less WPM than a Real EXTRA

K3UIM

  • Member
  • Posts: 2145
Re: 40M QRP amp from 807s?
« Reply #11 on: September 23, 2021, 07:08:30 AM »

RZP: "BTW, the frequency is 7MHz, not 7mhz ....mHz is millihertz -, one thousandth of a Hertz.....The difference may be obvious here, but it isn't  always!"

Or, some of us old fogies like to say, "7 Megs".  (Not politically correct.) <Hidden Text>disgusting, sound heard of breaking wind<End text>

 ::) ;D 8)

Charlie
« Last Edit: September 23, 2021, 07:13:15 AM by K3UIM »
Logged
Charlie. K3UIM
Where you are: I was!
Where I am: You will be!
So be nice to us old fogies!!

AC9QC

  • Posts: 57
    • HomeURL
Re: 40M QRP amp from 807s?
« Reply #12 on: September 29, 2021, 04:25:03 AM »

So I'm trying to determine if neutralization is going to be required for this. Looking at this reference I find it interesting in that it says of a push pull amplifier you need to run capacitors from the plate to the control grid of the opposite tube and that these values should match Cgp. In the case of the 807 the data sheet lists it as .2pF. I'm taking that as I'm unlikely to need neutralization at 30 megs and below? I mean I have some trimmers if needed that I could use however they're not near a fraction of a pico farad. I seem to recall something about the beam plates and screen grids having an effect on Cgp for some reason but it's been a while.

http://www.vias.org/basicradio/basic_radio_28_04.html

https://tube-data.com/sheets/049/8/807.pdf
Logged

KM1H

  • Member
  • Posts: 11155
Re: 40M QRP amp from 807s?
« Reply #13 on: September 29, 2021, 08:45:39 AM »

The tube and surrounding circuitry/metal affect that C value. At 807 voltages even mini Teflon cable is OK.  Ive also made several for 3-4 KV QRO amps out of larger coax such as real RG-8/213/214 and not a foam version.

Carl
Logged

G3RZP

  • Member
  • Posts: 2254
Re: 40M QRP amp from 807s?
« Reply #14 on: September 30, 2021, 03:15:42 AM »

The easy way is single ended and a Bruene capacity bridge neutralisation circuit. In this, the 'cold' end of the grid tuned circuit connects to the cathode via a capacitor, typically 500 to 1000pf. The neutralising capacitor, typically a 10 or 15 pF high voltage variable, connects from the plate to the junction of the 'cold' end of the grid  tuned circuit and the 500 or 1000pf capacitor that goes from tuned circuit to cathode. Single ended is also much easier for use with a pi network tank circuit, which is why push-pull PA stages pretty well disappeared at HF in the late 1940s when TV Interference became more of a problem because of harmonics - with which the pi tank helped.

Unless the layout is very poor, Class C 807s don't generally need neutralising below 30 MHz. Having said which, the WW2 USN and USAAF 'Command' transmitters neutralised the 1625s (12 volt heater 807s) even in the model going down to 2.1 MHz. The Command transmitters, the Collins TCS of the same era, the WW2 Bendix TA12 all used parallel 1625s because that was much easier circuit and component wise than push pull. Even the Marconi Marine 'Oceanspan' (which used three 807s) ran them in parallel and went up to 22MHz. One useful trick is to mount the tube poking through a hole in the chassis, with the base far enough down below the chassis that the chassis is level with lowest plate support mica in the tube. Some brands of 807 had the reputation of being prone to VHF parasitic oscillations: not a problem i've ever had, but I've always used something like 47 or 68 ohm non-inductive resistors with very short leads right up at the grid and screen grid pins at tube socket as well as the usual parallel resistor/inductor arrangement at the plate cap. Running as a linear may require neutralising on the higher bands; again, parallel operation is easier if you need to add neutralisation at a later stage.

The main reason for using push-pull in the Marconi tx I mentioned  was that the output capacity of the banks of 3 tubes was effectively in series with that of the other bank of three tubes: for export, especially in post WW2 conditions, the fewer different spare tube types, the more attractive the equipment!
Logged
Pages: [1] 2 3   Go Up