Texas Star DX500 and high SWR

[Shockwave]

Getting to all of that in the next post... like I said, we're handling one thing at a time. But remember, even after the chokes, you have RF bypass caps, and also the wirewound resistors are inductive, and act as chokes themselves.

Also, I want to make it very clear by what I meant, I think something I said was misunderstood..... if you read my previous post closely, I mentioned:

"The ONLY thing that would effect SWR in, or out, in an arrangement like this, is that if the bias supply changes, it can and will affect the input and output impedance of the transistors a little, which would cause the INPUT SWR to change a bit. It would also detune the output a little as well."

When I said "in an arrangement like this" I was referring to the bias circuit, and what I meant is, that everyone blaming the bias circuit for the input and output SWR issues is a long shot, and while it CAN affect the input and output impedances, it is, as I previously mentioned, a little change.

Bias in a high level stage like this will have very little effect on VSWR as you mention since the RF drive is already enough to forward bias the transistors to the point where impedance won't change much between class AB or class C. Oscillations can drive the input and output VSWR well into the red.

Or worded yet another way would be: In a longshot, if the bias supply failed, it is not the main cause, the end all, be all of the issues, so please quit randomly changing wirewound resistors and chokes trying to solve the issue.

You might want to disconnect R21 and R28 to shut the contaminated bias down and test your theory. I should have figured this out years ago because when the RF actually burns open the 10 uh choke, the keying circuit stops working. The interesting part is you can manually engage the relay without the high VSWR problem because that shuts the bias off just like lifting the two resistors will.

We also established that these devices have very little gain, if any at 54Mhz, and even less on odd order harmonics.

If you're talking about the finals, it is worth noting that the 2SC2879 only has a slight reduction in gain at 54 Mhz. It behaves very much like an MRF-492 on VHF lowband. I dropped four 2SC2879's in and old Motorola Micor crystaled up on 6 meters and it's been making 250 watts carrier with no problem since 2004.

Oh one more thing... you said that the bias is tied right to the input. Once again, not really. You have to look at the path. It has to go through the choke, then the resistor, 25 wirewound in this case, the leads that have ferrite over them, the input transformer which the way it is set up acts like a choke itself, also a lot of it is shunted via a bypass cap, and what is left over still has a 1.6 ohm resistor, which is part of the divider, and two 10 ohm resistors at the base of each transistor.

Even with all that filtering, at least 25% of these amplifiers still fail to run unconditionally stable because it's not good enough to remove 99.9% of the output RF from the input side. Any small changes in load reactance can easily shift the amp back into a condition where it will oscillate.

When you said that poor isolation is the cause of self oscillation, I would say...yes and no... meaning that you WANT a small amount of feedback, but negative feedback. Old solid state amplifiers would actually purposely take a small lead, insulated and wrap it around the input lead. This was tied to part of the ouput circuit, and sometimes 180 degrees out of phase. Crazy right? Not really. It was to stop self oscillation.

The key here is in these cases they ONLY add negative feedback by choosing a value of cap that will cause the 180 degree phase shift between current and voltage. It's the same principle as neutralizing a tube amp and never applies the positive feedback we see in this circuit.

But on this design, it does have negative feedback. Also, weather on purpose or on mistake, RF is routed through the power switch, so once again, purposely, or accidentally because they are routed closely, you get some feedback there. Also the 330pf capacitors help with any out of band oscillations as well.

Even if you push the amplifier near class A, where small signal gain will increase substantially, oscillation isn't an issue, unless something else is wrong.

But all of this can be verified with a spectrum analysis and sweep for in band spurs, oscillations, and while we are at it, check IMD levels (also in band) then check 3rd order harmonic levels.

I've already confirmed the problem in these amps is related to oscillations and have seen them on the spectrum analyzer and scope as described. It just took me until recently to figure out the cause looking at the schematic. IMD and 3rd order harmonics suck because they don't run the device at 60 watts output where the specs are good.

Ok, going back to the bias. I agree 100% that we need to change this. And we will...we are getting to that. We will use the "contaminated" bias supply to power the relay's coil, and use RF bypassing for good measure, since wiring the second relay in parallel with the other relay's coil will cause too much current across the keying transistor.

The 2N2907A keying transistor can support 600 ma of drive and be loaded down to 20 ohms at 12 volts. The 2 relay coils and meter light will still be well above 200 ohms.

But first, we are going to correct the heart of the issue, and that is improper impedance matching.

More tomorrow.

Adding a tuned input or output circuit may hide the problem since they are narrow band and could reject the oscillation frequency if it were far enough away from the fundamental frequency. If the circuit were designed correctly it would not need any of this to run stable although it would improve spectral purity and efficiency.

 

[2RT307]

Glad to see you guys are looking at this problem again. The tech aspect of it is way beyond my scope, but I'm certainly looking forward to someone finding a solution.

73,
RT307

 

[RadioGuy772]

" Oscillations can drive the input and output VSWR well into the red."

Agreed... for a few reasons, one is it can throw the input SWR all over the place, and even damage the exciter by feeding some weird, high power frequency that is unknown back into it. I have seen that happen before.

Also, I have seen many amplifiers that are going this, at some strange frequency, and output SWR is crazy as well, not into a dummy load, either because the dummy load is not frequency discrimitive like an antenna system, or perhaps because the dummy load is a purely resistive load and therefore tames the amplifier..right?

Before we go farther...do we agree on the above? I want to, and enjot going back and forth...not to debate, per se, but to learn, and bounce ideas off of each other... this is great!

Ok next..

"If you're talking about the finals, it is worth noting that the 2SC2879 only has a slight reduction in gain at 54 Mhz. It behaves very much like an MRF-492 on VHF lowband. I dropped four 2SC2879's in and old Motorola Micor crystaled up on 6 meters and it's been making 250 watts carrier with no problem since 2004."

Interesting... never actually tried it to be honest, just going by the spec sheets, although, wouldn't the transformers, ferrite mix chosen, and all other things around the design perform horribly at 54mhz, not to mention the push pull design, not to mention the 3rd harmonic being even less? I am still have trouble buying this, because years ago, we did to testing using an HP Spectrum Analyzer. At least on the output (SO-239) figures weren't great, but acceptable, when drive level was kept reasonable.

"The key here is in these cases they ONLY add negative feedback by choosing a value of cap that will cause the 180 degree phase shift between current and voltage. It's the same principle as neutralizing a tube amp and never applies the positive feedback we see in this circuit."

Ok gotcha. But not sure where what we are stating differs here?

"The 2N2907A keying transistor can support 600 ma of drive and be loaded down to 20 ohms at 12 volts. The 2 relay coils and meter light will still be well above 200 ohms."

Still choosing not to do this, because the transistor's maximum could be exceeded by someone using a "who knows what" relay, with a "Who knows what" coil. Then we would hear crying about ho after they applied the mod, their "keying" transistor keeps going out.

"Adding a tuned input or output circuit may hide the problem since they are narrow band and could reject the oscillation frequency if it were far enough away from the fundamental frequency. If the circuit were designed correctly it would not need any of this to run stable although it would improve spectral purity and efficiency."

Yes, it very well could "hide" it, although by raising the Q just a bit, and also just focusing on impedance matching, we are in fact increasing efficiency, lowering IMD, and not having the amplifier so touchy to changes in the output load, while also producing a stable load to the exciter on the input... those are all good things.

 

[Robb]

Not quite clear on the source of these oscillations you are talking about.

IS this because of the circuit not having enough adjacent rejection/filtering, a radio being used with spurious emissions, amp design leaking RF where it shouldn't be/improper bypassing, or just being plain over-driven?

Still trying to figure amps out - thnx . . .

 

[Shockwave]

" Oscillations can drive the input and output VSWR well into the red."

Agreed... for a few reasons, one is it can throw the input SWR all over the place, and even damage the exciter by feeding some weird, high power frequency that is unknown back into it. I have seen that happen before.

Those are the worst case scenarios with self oscillations. I seen a guy destroy a $5,000 Yaesu when a tetrode amp self oscillated while tuning it.

Also, I have seen many amplifiers that are going this, at some strange frequency, and output SWR is crazy as well, not into a dummy load, either because the dummy load is not frequency discrimitive like an antenna system, or perhaps because the dummy load is a purely resistive load and therefore tames the amplifier..right?

Before we go farther...do we agree on the above? I want to, and enjot going back and forth...not to debate, per se, but to learn, and bounce ideas off of each other... this is great!

We are in absolute agreement here.

Ok next..

"If you're talking about the finals, it is worth noting that the 2SC2879 only has a slight reduction in gain at 54 Mhz. It behaves very much like an MRF-492 on VHF lowband. I dropped four 2SC2879's in and old Motorola Micor crystaled up on 6 meters and it's been making 250 watts carrier with no problem since 2004."

Interesting... never actually tried it to be honest, just going by the spec sheets, although, wouldn't the transformers, ferrite mix chosen, and all other things around the design perform horribly at 54mhz, not to mention the push pull design, not to mention the 3rd harmonic being even less? I am still have trouble buying this, because years ago, we did to testing using an HP Spectrum Analyzer. At least on the output (SO-239) figures weren't great, but acceptable, when drive level was kept reasonable.

Motorola has been using push pull solid state PA sections in all low band radios above 60 watts since the early 1970's. Of course their core material in the transformers and combiners is optimized for this range where most CB stuff uses the 2 to 30 Mhz mix. When I placed four 2SC2879's in the VHF low band Micor it actually had more gain and output than the old school stock transistors.

"The key here is in these cases they ONLY add negative feedback by choosing a value of cap that will cause the 180 degree phase shift between current and voltage. It's the same principle as neutralizing a tube amp and never applies the positive feedback we see in this circuit."

Ok gotcha. But not sure where what we are stating differs here?

Just pointing out that the RF feedback through the bias circuit is not inverted 180 degrees and will and therefore contribute to self oscillations even though the circuit uses an additional negative feedback loop.

"The 2N2907A keying transistor can support 600 ma of drive and be loaded down to 20 ohms at 12 volts. The 2 relay coils and meter light will still be well above 200 ohms."

Still choosing not to do this, because the transistor's maximum could be exceeded by someone using a "who knows what" relay, with a "Who knows what" coil. Then we would hear crying about ho after they applied the mod, their "keying" transistor keeps going out.

My suggestion would actually lower the current through the keying transistor since it will no longer be providing the forward bias current. Radio Shack sells a SPST relay with a 12 volt 750 ohm coil.

"Adding a tuned input or output circuit may hide the problem since they are narrow band and could reject the oscillation frequency if it were far enough away from the fundamental frequency. If the circuit were designed correctly it would not need any of this to run stable although it would improve spectral purity and efficiency."

Yes, it very well could "hide" it, although by raising the Q just a bit, and also just focusing on impedance matching, we are in fact increasing efficiency, lowering IMD, and not having the amplifier so touchy to changes in the output load, while also producing a stable load to the exciter on the input... those are all good things.

All of those ideas are good design practice however, a properly designed broad band RF amp will not need tuned circuits to run stable. They are added to improve the match and reduce harmonics. The problem with this amp is RF getting into the bias circuit. The proof here is when that 10 uh choke burns open and shuts down the keying circuit AND bias, the amp runs clean and stable if you key the relay by hand.

 

[Shockwave]

Not quite clear on the source of these oscillations you are talking about.

IS this because of the circuit not having enough adjacent rejection/filtering, a radio being used with spurious emissions, amp design leaking RF where it shouldn't be/improper bypassing, or just being plain over-driven?

Still trying to figure amps out - thnx . . .

Self oscillations can occur for many different reasons all related to in phase output power being coupled back to the input stage. This can happen from poor RF layout such as Texas Star's wiring all the RF circuits through the front panel push buttons. It can be caused by not properly decoupling the DC power source from RF. In this case it's caused because bias is switched on through the same relay contact carrying the RF output to the antenna.

This is a very poor design with no understanding of how much attenuation is really required to make the circuit run stable. Also worth considering is what happens on SSB with the bias controlled by the keying transistor. On strong audio peaks the keying transistor is fully conducting and supplying the maximum bias current. With softer modulation on SSB the keying transistor will not maintain full conduction and reduce bias exactly when we need it the most.

 

[Robb]

Self oscillations can occur for many different reasons all related to in phase output power being coupled back to the input stage. This can happen from poor RF layout such as Texas Star's wiring all the RF circuits through the front panel push buttons. It can be caused by not properly decoupling the DC power source from RF. In this case it's caused because bias is switched on through the same relay contact carrying the RF output to the antenna.

This is a very poor design with no understanding of how much attenuation is really required to make the circuit run stable. Also worth considering is what happens on SSB with the bias controlled by the keying transistor. On strong audio peaks the keying transistor is fully conducting and supplying the maximum bias current. With softer modulation on SSB the keying transistor will not maintain full conduction and reduce bias exactly when we need it the most.

For those of use to do a lot of SSB, this comes as a bit of a shock. What can be done to remedy this bias/SSB issue before it causes failures? Can just lowering the radio's ALC level and thereby reducing the input power suffice; or is a modification/re-design necessary?

 

[RadioGuy772]

And to think we haven't even *touched* the subject of the unregulated bias that goes all over the place when excitation (drive) is applied at the base of the transistors.

A bias source, needs to be a stable, low impedance source, and be capable of the current required.

We can still go step by step, which is kind of how I anted to do things, BUT, for the sake of argument, we'll throw a "spoiler" in there for those that want to just get to the point.

My solution, or what I have done, is change the input circuit, bypassing the switches, variable, resistors, etc, added some capacitance to each input transformer and changed the value of the capacitor that goes from the input combiner to circuit ground, and the result is a very low input SWR, and the power came right up.

On the output, I ususally add a little capacitance, usually I just remove what they have there, and use a 1200pf metal mica on each output transformer. I do away with the capacitor in the output stage that basically just blocks the DC from the so called bias supply, since it on't be needed anymore. I also change the value of the capacitor that goes from the output combiner to circuit ground as well to "match" things up a bit more closely.

Next, I add a small transistor that the 2n2907a will switch, that will switch the original T/R relay, and a seond relay, remember to take into account the "SSB delay" and also adding anti-spike diodes across the relays coils.

Then the second relay turns on and off the VCC to a couple of 7809 regulators, and circuitry is added that is needed when you are using two regulators. (don't just connect them together) those go to a pass transistors of your choice, and after some RF bypassing, a few electrolytics to stiffen up the source a little, etc, you have your new, switched bias source. I use one diode across each set of transistors (2 total for 4 transistors) with some thermal compound for temp tracking..a little crude but it works.

I set EACH 2SC2879 transistor to 140ma of idle current. YES I KNOW..a lot of people recommend that or less for a set. But if things are matched properly, cooled properly, and not driven far beyond compression, I have not had any failures.

Because the small signal gain will go up when doing this, you do need to make sure you have everything right, since we were talking about oscillations here. BUT, usually the standard 100 ohm / .01 arrangement works here, although, when looking at spectral purity, for some reason I get better results using about 82 ohms. Yes, I know, more neg feedback, right? But 68 makes it worse. 82 seems perfect, so it's what I always go with.

At 14.3 volts, with 50 watts PEP in, I have no problems seeing well over 600 watts output, no more nigh SWR issues, and while it doesn't meet commercial specs, the bias remains stable, and output is fairly clean.

I was going to take a step by step approach to all of this, giving component values, and ways to do the changes, but I think we all got ahead of it. I still can, and probably will, but before doing so.. maybe I could get some input, and feedback (no pun intended) on some of this, and maybe see what people think.

 

[Shockwave]

For those of use to do a lot of SSB, this comes as a bit of a shock. What can be done to remedy this bias/SSB issue before it causes failures? Can just lowering the radio's ALC level and thereby reducing the input power suffice; or is a modification/re-design necessary?

Well, I may have made the problem sound worse than it is because the more serious issue is RF getting into the DC bias and applied back to the base of the transistors. The voltage problem with the SSB bias controlled through the keying transistor is somewhat smoothed out by the SSB delay and the added electrolytic cap.

You will notice on SSB when you pause between words or speak softly, the meter light will dim just before it goes out as the cap discharges. By adding the relay you provide a hard switch for the bias where it's either completely on or off. None of this in-between stuff.

The bias circuit in these amps is perhaps one of the worse attempts I've ever seen using a resistive divider. There is nothing whatsoever in this circuit that could stabilize the voltage. Any percentage of change in the supply voltage will cause the same percentage of change in the bias.

At the very least the 1.8 ohm resistors should be replaced or paralleled with silicon diodes heat sink compounded to the top of the transistor case. This will help regulate the voltage and provide thermal tracking. Even better is to add a 3 terminal regulator as the source for the bias voltage.

 

[RadioGuy772]

One more thing, on a side note, that got me "fired up" about this whole subject, is that I recently had a conversation with a couple folks down there at texas star / galaxy / RCI and brought several design flaws to their attention, such as small changes in the amps, and receiver issues as well. Not anything that would break the bank, just little changes. What I got back verbally was that "We can't duplicate the problems, and we have been using the same design for years".

I brought up the fact (off suject) that the receivers have entirely too much gain in the front end, wide as heck filtering, issues in the IF section, traces in their 2950DX series radios ran too close together that act as capacitors....just a lot of things. I think the attitude is, that people keep buying the products, so why should we change it.

EDIT: Also, I wanted to thank you, Shockwave for all of your time, knowledge, and insight into these issues. I really enjoy learning and also sharing my knowledge with people. It keeps things interesting.

 

[Shockwave]

And to think we haven't even *touched* the subject of the unregulated bias that goes all over the place when excitation (drive) is applied at the base of the transistors.

A bias source, needs to be a stable, low impedance source, and be capable of the current required.

We can still go step by step, which is kind of how I anted to do things, BUT, for the sake of argument, we'll throw a "spoiler" in there for those that want to just get to the point.

My solution, or what I have done, is change the input circuit, bypassing the switches, variable, resistors, etc, added some capacitance to each input transformer and changed the value of the capacitor that goes from the input combiner to circuit ground, and the result is a very low input SWR, and the power came right up.

On the output, I ususally add a little capacitance, usually I just remove what they have there, and use a 1200pf metal mica on each output transformer. I do away with the capacitor in the output stage that basically just blocks the DC from the so called bias supply, since it on't be needed anymore. I also change the value of the capacitor that goes from the output combiner to circuit ground as well to "match" things up a bit more closely.

Next, I add a small transistor that the 2n2907a will switch, that will switch the original T/R relay, and a seond relay, remember to take into account the "SSB delay" and also adding anti-spike diodes across the relays coils.

Then the second relay turns on and off the VCC to a couple of 7809 regulators, and circuitry is added that is needed when you are using two regulators. (don't just connect them together) those go to a pass transistors of your choice, and after some RF bypassing, a few electrolytics to stiffen up the source a little, etc, you have your new, switched bias source. I use one diode across each set of transistors (2 total for 4 transistors) with some thermal compound for temp tracking..a little crude but it works.

I set EACH 2SC2879 transistor to 140ma of idle current. YES I KNOW..a lot of people recommend that or less for a set. But if things are matched properly, cooled properly, and not driven far beyond compression, I have not had any failures.

Because the small signal gain will go up when doing this, you do need to make sure you have everything right, since we were talking about oscillations here. BUT, usually the standard 100 ohm / .01 arrangement works here, although, when looking at spectral purity, for some reason I get better results using about 82 ohms. Yes, I know, more neg feedback, right? But 68 makes it worse. 82 seems perfect, so it's what I always go with.

At 14.3 volts, with 50 watts PEP in, I have no problems seeing well over 600 watts output, no more nigh SWR issues, and while it doesn't meet commercial specs, the bias remains stable, and output is fairly clean.

I was going to take a step by step approach to all of this, giving component values, and ways to do the changes, but I think we all got ahead of it. I still can, and probably will, but before doing so.. maybe I could get some input, and feedback (no pun intended) on some of this, and maybe see what people think.

It actually sounds like a job well done. As you know it's nearly impossible to maintain a perfect 50 ohm input impedance through all that crap they pass the drive trough. How many ohm impedance are those front panel switches? All the exposed unshielded traces carrying both input and output RF around those switches. Bypassing that crap is a good idea.

With respect to the 1200 PF metal caps across the collectors, you just might want to check that on word peaks, the maximum current drain always remains below 50 amps. With that much loading on the push pull bank you may be approaching maximum rated current on the finals.

Each pair of push pull 2SC2879's can draw as much as 25 amps. Even a brief excursion beyond this point can destroy the transistors instantly. If you find you're close to 50 amps now, using 1000 pf caps across the collectors will reduce the current drain with only a minor loss in PEP output.

Bias may be a bit high but if you work lots of SSB and have regulated it to this degree, you should be fine. It may dissipate some extra heat as a result so a fan is a good idea if you're long winded. With respect to the negative feedback resistor value, everything from the components to the layout can effect the optimal value. Too many ohms and you don't provide enough negative feedback to stabilize. Too little ohms and you destroy the gain.

You may have several hours labor involved with all these improvements and seem pleased with the results of your work. Many don't have the ability to take it this far or care to pay for those hours of labor. The average customer just knows when he turns his Texas Star on the perfect VSWR just went up to 3:1 and no tuning of the antenna can correct it.

They usually like to pay for ONLY what they need but not less than required to get the job done right the first time. They are willing to accept a less than perfect input VSWR with some of the settings provided on the front panel so long as it doesn't upset the match on the output side.

This can be accomplished simply by isolating the bias from being switched through the RF output contact on the relay. The usefulness of your extra modifications are only apparent to the knowledgeable user. Someone who understands "working" may not be good enough to reproduce the quality they demand.

 

[RadioGuy772]

Thank you for the "job well done" and also all of your suggestions and input.

I really do appreciate it...VERY much!

Maybe next is a write up on how to do everything, step by step, section by section. I don't mind doing this, the only problem, is that when I have done this in the past, such as changing audio circuits to change AF response, etc, I would find them a year later plastered or changed on the internet and would be incorrect, or someone would pick and choose the steps and only do a portion, not understanding that a change on one side of the board affects something else on the other side. But I agree, in fact, I was on the phone with another forum member I have knows for a while last night discussing exactly what you said about the double sided board, poor layout, traces routed wrong, and finally RF through push button switches! Someone must have had a million of those things around. I can see no other logical explanation on why they would do this. Imagine if we were working with VHF! No ground plane construction, impedance upsets everywhere, etc. We would have to change the subject from "amplifier" to maybe "attenuator".

But back to the point, there are fixes that can be done, and it can be made to work decently.

 

[Shockwave]

One more thing, on a side note, that got me "fired up" about this whole subject, is that I recently had a conversation with a couple folks down there at texas star / galaxy / RCI and brought several design flaws to their attention, such as small changes in the amps, and receiver issues as well. Not anything that would break the bank, just little changes. What I got back verbally was that "We can't duplicate the problems, and we have been using the same design for years".

I brought up the fact (off suject) that the receivers have entirely too much gain in the front end, wide as heck filtering, issues in the IF section, traces in their 2950DX series radios ran too close together that act as capacitors....just a lot of things. I think the attitude is, that people keep buying the products, so why should we change it.

EDIT: Also, I wanted to thank you, Shockwave for all of your time, knowledge, and insight into these issues. I really enjoy learning and also sharing my knowledge with people. It keeps things interesting.

Sounds like the one I had with Yaesu a few years ago. More and more frequently we find these manufacturers either play dumb or don't even hide their lack of concern when they make obvious design errors. If it could cost them money, they will fight tooth and nail to steer you in another direction.

Everything you say about the main board used in those "export" radios is true. They never fix any of the old problems as they "update" but just continue to add more new problems to go after. How do you like the tone quality out of the 2950 receiver? My Grandmother has a crank up Victrola that reproduces the same fidelity and static pops.

Nothing but piercing mid range and treble no matter what speaker you put on them. The models that have a tone control only cut the treble and do nothing to add any sign of bass. Turns out the value of one ceramic coupling cap off the detector is far to low in value. Changing the value brings the receiver fidelity to life with full tone and would cost far less than adding the useless tone potentiometer that can't restore the low end.

The gain they run the IF strip at is way too high because they tried to compensate for the lack of the bandwidth in the RF strip to cover 5 or 6 Mhz. The result is way too much gain in the center of the bandwidth through the RF strip and total loss of selectivity in the IF strip. When the radio ALWAYS shows a noise level on the S meter, they ran the gain up too high. Turning the RF gain down doesn't have the same effect as getting the gain right in the IF strip either.

You're quite welcome for any responses. I enjoy seeing others like yourself arrive on the scene able to contribute as well. Happy New Year!

 

[RT2020]

So what is the usual fee for a fix on these amplifiers? I assume the 667 wold be the same.

 

[Shockwave]

So what is the usual fee for a fix on these amplifiers? I assume the 667 wold be the same.

Qualified shops should be able to stabilize a Texas Star for an hours worth of labor and few parts. That would just be addressing the basic problems associated with RF feedback that can cause high VSWR. At this point the amp will perform up to most peoples expectations. It will sound relatively clean but won't be wining any awards for spectral purity.

The extra steps RadioGuy has done with his can be rather time consuming. Regulating the bias, adding thermal tracking, tuned circuits on both the input and output, tweaking the match on the transformers and padder caps to perfect the match, adding bypassing caps to decouple the amp from the DC supply (and the DC feed to the driver in the 667) all add up to more labor and parts.