View Full Version : Handbook dual 3-500Z design
OK, so I'm looking through some of my old handbooks and read through the dual 3-500Z amp project in the 94 edition.
The design looks pretty standard and the write up claims it'll do 1500W out.
Problem is it never really gets there. If you look at the test specs with 3500V on the plates and 100W input it almost makes 1500W out on 80 and 40, drops to 1300W on 20 and by the time we get to 10m power out is 900W.
That's a 500W drop in PO from 80 to 10m.
What's the cause?
What's the cause?
Poor design, input alignment, driver rig power foldback, or soft tubes.
A SB-220 or L-4B has less variance than that. I also do not like 3500V on that tube, especially Chinese but Eimacs also will object forcefully.
The B&W PT-2500A ran 3600V and would run 1700-1800W 160-20M and 1500 by the time it got to 10M. The AL-82 isnt bad either.
Any 3-500 will have gas arcs especially Chinese due to insufficient processing. Testing by US sellers is done at a much reduced voltage for "matching" or catastrophic failure weeding out only. They will arc sooner at 3500V than 2500V and the only way to prevent it is to run the anodes a good bright red on tantalum plates or a darker brick red with graphite on a regular basis. The gettering material is on the anode of both type.
Carl
Poor design, input alignment, driver rig power foldback, or soft tubes.
A SB-220 or L-4B has less variance than that. I also do not like 3500V on that tube, especially Chinese but Eimacs also will object forcefully.
The B&W PT-2500A ran 3600V and would run 1700-1800W 160-20M and 1500 by the time it got to 10M. The AL-82 isnt bad either.
Any 3-500 will have gas arcs especially Chinese due to insufficient processing. Testing by US sellers is done at a much reduced voltage for "matching" or catastrophic failure weeding out only. They will arc sooner at 3500V than 2500V and the only way to prevent it is to run the anodes a good bright red on tantalum plates or a darker brick red with graphite on a regular basis. The gettering material is on the anode of both type.
Carl
Kinda what I figured. I suspect the tank circuit they used isn't quite up to snuff because the efficiency really takes a dive..
The amp was designed to run with 3 KV on the plates. The output was no better than 1300W with 3 KV.
3.5K was available for soft mains and it was the only way to get close to 1.5KW out.
AF6LJ
04-26-2011, 01:04 AM
Poor output circuit design and layout could be the culprit.
Interesting comments but not realistic. A lowly Ameritron AL80B will output 1kw peak using one 3-500z tube with 3kv on the plate. Why wouldn't two tubes put out just 50% more? I have tested that amp and it will put out 1kw peak but you have to drive the plate current well over the 400ma limit per the Eimac data sheet.
I have several 3-500z amps and all easily put out 1.5kw with only 2800v on the plate with about 110 watts drive.
That's a 10-4 :D
Interesting comments but not realistic. A lowly Ameritron AL80B will output 1kw peak using one 3-500z tube with 3kv on the plate. Why wouldn't two tubes put out just 50% more? I have tested that amp and it will put out 1kw peak but you have to drive the plate current well over the 400ma limit per the Eimac data sheet.
I have several 3-500z amps and all easily put out 1.5kw with only 2800v on the plate with about 110 watts drive.
I agree with you.
I brought it up because this was a project in the ARRL handbook for several years.
I'm wondering what causes the precipitous drop in efficiency as you move up in frequency, and the lack of actually getting 1500W out, even with 3.5K on the plates.
Seems to me the loaded Q of the tank circuit on the higher bands must be sky high which could be caused by getting the optimum plate load resistance wrong.
I just perused the operating manual for the Ameritron AL-82.
It runs with a full load voltage of 3.3 KV on the plates.
Assuming the Ameritron numbers of 65% efficiency (CW) you need about 2307W input to get 1500W out.
2307/3300 = .699ma plate current or right at 350ma per tube. Pretty close to the 400ma max per tube.
Dissipation is 807W or a smidgen (technical term) over 400W (500W max) per tube.
Another possibility is the power supply is sagging badly under load along with the inefficient tank.
OK, so I'm looking through some of my old handbooks and read through the dual 3-500Z amp project in the 94 edition.
The design looks pretty standard and the write up claims it'll do 1500W out.
Problem is it never really gets there. If you look at the test specs with 3500V on the plates and 100W input it almost makes 1500W out on 80 and 40, drops to 1300W on 20 and by the time we get to 10m power out is 900W.
That's a 500W drop in PO from 80 to 10m.
What's the cause?
It can't be voltage or tubes, because it works on 80. :-)
I don't know why so many people leap at the tank!! It doesn't matter significantly if tank loaded Q is 25 or 12, output will be about the same. It is almost never the tank if the tank is not melting down and is hitting resonance. Most commonly, unless they used some nichrome or some other silly thing in the anodes, a drop in power like that is caused by a bad input circuit design IF it is accompanied by a loss in efficiency.
What happens (in grounded grid amps) is the cathode is in SERIES with the output. The cathode has all the same harmonics as the anode system, because they are both in series.
If the design is such that the cathode has a high impedance to ground at harmonics, in particular even harmonics, the waveform on the anode gets very "soft" switching in and out of conduction. The tube hangs in the area where anode to ground resistance is dissipating energy much longer than a tube with low impedances in the cathode at harmonics, and efficiency goes to pot.
I first found this in a Heath amp I designed (it was supposed to be called the Warrior II, but Heath pulled out of the kit business before release). It made great efficiency and power on 160-20 and then started quickly going to heck on 15 and up. On ten meters it made about 25-30% efficiency.
The problem was the length of the coax from input circuit to the tube cathodes. I had to relocate the input circuit and change the configuration, and the efficiency went up to about 60% on ten meters.
The same thing happened in the AL12 series of Ameritron amps, a totally different layout than the Warrior II. In the AL12 series changing the coax between the tube and the tuned input from 50 ohms down to 16-25 ohm impedance cures the problem.
That aside, if the amp you are referring to is the one I am thinking of it has a multitude of shortfalls...some dealing with safety.
73 Tom
Thanks Tom.
I hadn't even considered that it might be the input side causing the problem.
Now that I think about it, it makes sense.
I suppose if one wants to home brew an 2x 3-500Z amp the AL-82 would be a good place to start.
Thanks Tom.
I hadn't even considered that it might be the input side causing the problem.
Now that I think about it, it makes sense.
I suppose if one wants to home brew an 2x 3-500Z amp the AL-82 would be a good place to start.
Let's look at it logically. The amp made almost 1500 watts on 80. That rules out power supply and everything else except the tank and input system. There is 30% missing on ten meters, 500 watts.
The typical small tank only looses about 1-5% of RF power. The more power the amp is designed for, the less loss as a percentage in the tank because components and wiring are larger. If we doubled that by halving unloaded Q or doubling loaded Q, we'd only lose 2-10%.
If the tank is hitting resonance and is not melting down in a few seconds, it cannot be the tank. As a matter of fact it almost never is a tank causing problems like that.
So we are left with the input circuit and the parasitic suppression. Nichrome suppressors have very low HF Q (but the same basic VHF Q, so they don't help there), and can really kill efficiency on ten meters. But again they would be red hot with 500 watts of heat. Even if nichrome is used, it will only cost about 100-200 watts out of 1500.
This pretty much takes it back to input circuit problems.
In nearly all cases of fresh grounded grid designs I have seen produce abnormally low efficiency on upper bands, it has been the input circuit. It has specifically been the impedance presented to the cathodes by the input circuit.
The tank makes little sense at all if it is hitting resonance and is not smoking or glowing red.
73 Tom
I hadn't even considered that it might be the input side causing the problem.
Now that I think about it, it makes sense.
As I said way back in reply #2:
Poor design, input alignment, driver rig power foldback, or soft tubes.
Carl
AF6LJ
04-27-2011, 03:44 PM
To paraphrase what Tom said the input is effectively in series with the output.
This lies at the crux of why I really don't care for grounded grid amplifiers.
There isn't the type or degree of isolation between input and output like there is in a grid driven amplifier.
While I maybe nit picking.......
It wasn't mentioned above as to the status of the plate tuning on the amplifier in question
Was it in resonance and properly loaded?
What was the grid current at full output on all bands.
Without sufficient information it's easy to get stuck on what could be the obvious problem when there is a lack of data.
With that said;
You can always count on Tom for what is most likely what is going on and to provide a clear explanation of same,
Thanks Tom. :)
To paraphrase what Tom said the input is effectively in series with the output.
This lies at the crux of why I really don't care for grounded grid amplifiers.
There isn't the type or degree of isolation between input and output like there is in a grid driven amplifier.
The output and input in series is a great thing Sue, if the grids are grounded and not half-floated. The grounded grid system generally has very high and very stable negative feedback. This stabilizes the amp and greatly reduces distortion. A typical grounded grid triode will often blow away a grid driven tetrode for odd-order intermodulation performance.
The only problems occur when the input circuit does not present a low impedance at the cathode for harmonics, when the grid lead is too long to be perfectly grounded, or when the design lifts the grid. Other than the lower gain caused by the negative feedback, GG amps are really good.
:-)
73 Tom
AF6LJ
04-27-2011, 10:06 PM
The output and input in series is a great thing Sue, if the grids are grounded and not half-floated. The grounded grid system generally has very high and very stable negative feedback. This stabilizes the amp and greatly reduces distortion. A typical grounded grid triode will often blow away a grid driven tetrode for odd-order intermodulation performance.
The only problems occur when the input circuit does not present a low impedance at the cathode for harmonics, when the grid lead is too long to be perfectly grounded, or when the design lifts the grid. Other than the lower gain caused by the negative feedback, GG amps are really good.
:-)
73 Tom
I get it Tom, but I have issues. with control over inverse feedback in order to improve linearity, and constant gain over all the HF bands.
With a grounded grid amp; what you see is what you get, trying to control gain and improve linearity is problematic.
Tubes with low lead inductance make grounded grid amplifiers work well.
811s, 572s, can work well on the low bands but I can't help but be concerned about the strays within those tubes and their effects on the higher bands (21, 28MHZ). At least with a 3-500Z there are parallel grid connections to help minimize the strays and a low inductance grid structure design, I can't say that for some of the other tubes. Linearity is still not easy to manage, you are stuck with no way to apply inverse feedback to level gain, and improve linearity.
To paraphrase what Tom said the input is effectively in series with the output.
This lies at the crux of why I really don't care for grounded grid amplifiers.
There isn't the type or degree of isolation between input and output like there is in a grid driven amplifier.
While I maybe nit picking.......
It wasn't mentioned above as to the status of the plate tuning on the amplifier in question
Was it in resonance and properly loaded?
What was the grid current at full output on all bands.
Without sufficient information it's easy to get stuck on what could be the obvious problem when there is a lack of data.
With that said;
You can always count on Tom for what is most likely what is going on and to provide a clear explanation of same,
Thanks Tom. :)
The only data provided in the handbook was Plate voltage, either 2.5, 3 or 3.5 KV, drive applied 100W with a couple bands at 110W, and Power out on each band.
I can only assume that the ARRL guys know the grid current limits for 3-500's and would design an amp with enough margin not to violate those limits.
The amp was built by Dick Stevens, now an SK.
If I am thinking of the right amplifier, it was full of design problems. Look to see if the filament transformer and/or blower have one side grounded to the chassis, and if it has hairpin nichrome suppressors. If it does, it is the circuit that has several issues.
Maybe it is a different one.
If I am thinking of the right amplifier, it was full of design problems. Look to see if the filament transformer and/or blower have one side grounded to the chassis, and if it has hairpin nichrome suppressors. If it does, it is the circuit that has several issues.
Maybe it is a different one.
Yep, sure does, one side of the filament transformer primary with a note explaining that chassis ground is not the same as AC ground.
3 turns of nichrome wire with a pair of 100 Ohm 2W resistors in parallel for suppressors.
The grid currents drawn are given for each band. At 3500V and 100-110W of drive they range from 180ma on 80 and 40 to 140ma on 10m.
The amp was loaded to the max plate current of 800ma on all bands for the tests.
The input network is a PI with fixed inductance wound on T-50-2 and -6 cores , one fixed cap (output) and the adjustable element was a mica compression trimmer (input).
Wonder why he didn't use slug tuned coils instead of a variable cap.
Obviously a very poor input circuit, even a SB-220 will run 250 ma Ig on all bands with 100W drive. Just a long coax can kill an otherwise good circuit. You can see the same thing with outboard input circuits on vintage amps that didnt come with any. The xcvr now sees a good match but the output isnt what was expected. Often the fix is as simple of subtracting the coax C from C-2 on the input board by reducing its value and retuning.
I knew the designer well as he lived in the same state, NH, and was always buying parts from me at Radio Kit, he would drive down for big orders to save on shipping:rolleyes:
Some of his stuff worked well, I still use a 4CX250 cavity for 440 FM for contests that was an 80's article.
Carl
Yep, sure does, one side of the filament transformer primary with a note explaining that chassis ground is not the same as AC ground.
3 turns of nichrome wire with a pair of 100 Ohm 2W resistors in parallel for suppressors.
The grid currents drawn are given for each band. At 3500V and 100-110W of drive they range from 180ma on 80 and 40 to 140ma on 10m.
The amp was loaded to the max plate current of 800ma on all bands for the tests.
The input network is a PI with fixed inductance wound on T-50-2 and -6 cores , one fixed cap (output) and the adjustable element was a mica compression trimmer (input).
Wonder why he didn't use slug tuned coils instead of a variable cap.
That is the amp with design issues.
I understand the variable caps in a homebrew one-off piece of gear. That might be easier than cut and try with a handful of micas. A toroid is adjustable because the low ui iron cores, contrary to some rumor, have flux leakage that changes mutual coupling as turns are spread or compressed. You can get about a 1.5:1 adjustment range in many small toroids by squeezing or spreading turns because of the flux leakage in the core.
The nichrome suppressors reduce HF Q, NOT VHF Q. The reduction of HF Q is largest around ten meters. The nichrome does exactly the opposite of what is proclaimed. The path through the inductor is primarily low frequencies, and the path through the shunt resistor is primarily at upper VHF. Changing the wire in the suppressor to nichrome reduces HF Q the largest amount, and has the very least effect on VHF. The maximum tank unloaded Q reduction caused by the suppressors is up on ten meters.
This, coupled with the poor input circuit location and other errors, is what that amp sucks on ten meters. It also is not safe as drawn. Low grid current is likely a combination of the impedance reflected back to the tube cathodes by the input design and the loss of tank Q at ten meter by the nichrome coils.
That design never should have made the Handbook.
73 Tom
I ran it mainly when I was going to college (late bloomer :D) and then just let it die off and went back to industry full time until retiring.
Ive been rekindling it recently primarily to sell off old stock and reinvent as strictly for amp parts.
