Electronic bias systems vs. PureSignal

User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Electronic bias systems vs. PureSignal

Postby w-u-2-o » Mon May 01, 2017 1:28 am

Possibly of interest to other Alpha 89 owners:

Recently I was speaking to another ham who had been having some trouble getting PureSignal to run in a stable fashion with his Alpha 89 and 200D. Make a long story short, apparently the Alpha has some sort of automatic bias control that would cause the auto-attenuate feature to be confused. Reverting to manual attenuation fixed the problem.

73!

Scott
User avatar
VY2WW
Posts: 15
Joined: Mon Apr 10, 2017 6:31 pm
Location: PEI

Re: Alpha 89 vs. PureSignal

Postby VY2WW » Tue May 02, 2017 11:05 am

A lesson learned with variable-bias linear amps like the 89 is to deactivate the DEXP setting to avoid generating brief transient bursts of IMD (easily visible on the transmit panagraphic display) when transitioning from zero db mic input. Even a little background noise, like the amplifier fan, is enough to avoid generating these visible initial spikes between pauses in speaking.

The 89 doesn't ship with a full schematic, and there's no mention of the bias circuit in the manual; anyone with good knowledge on how to turn OFF this bias control, please chime in. I believe there are some other linear amps using bias control circuitry that may present similar complications for PS operations, as well. I know the designers meant well, but with the evolution of SDRs and PS, well - unintended consequences!
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Alpha 89 vs. PureSignal

Postby w-u-2-o » Tue May 02, 2017 1:19 pm

It's not surprising that any amp that changes its bias state with different input conditions would cause problems with PureSignal. As PureSignal either executes auto-attenuate and or linearization states, if that causes a bias state change in the amp, then that could wind up being the worst kind of so-called "memory effect", as a change in bias state equates to a change in linearity. If the corresponding state changes represent an under damped or or critically damped condition, then the PureSignal solution, and the amp bias state, would oscillate all over the place.

Or, in simpler terms, the amp and PureSignal will chase each other's tails! ;)

73,

Scott
User avatar
w9ac
Posts: 290
Joined: Sun Apr 09, 2017 4:01 pm

Re: Alpha 89 vs. PureSignal

Postby w9ac » Tue May 02, 2017 1:21 pm

VY2WW wrote: The 89 doesn't ship with a full schematic, and there's no mention of the bias circuit in the manual; anyone with good knowledge on how to turn OFF this bias control, please chime in. I believe there are some other linear amps using bias control circuitry that may present similar complications for PS operations, as well. I know the designers meant well, but with the evolution of SDRs and PS, well - unintended consequences!


Concerning the Alpha 89, I would look to another root cause. ETO Alpha models 70, 77, 77D/S amps DO use RF switched EBS, as do the 91B, 99, 8100, 8406 and 8410 amps. As designed, EBS with RF input detection is used to save wasted heat when the tube(s) are idle between CW elements and pauses in speech. EBS can also invoke brief bursts of IMD when the tube is deeply cut off, then required to quickly change from Class C to AB. Many EBS systems are designed with far too much cut-off bias applied. Probably a better approach is to bias only slightly into Class C in standby, then quickly switch to AB with just a few mW of detected input power.

Alpha models 89, 87A and 9500 do NOT switch tube bias via RF detection, nor do they employ EBS. Those amps have two states: (1) tube(s) normally cut off when not in a Tx state; and (2) tube(s) turned on into conduction via the amp key line. The bias is only switched when the key line is closed, the amp is in the "operate" mode, and no faults are detected. Early amps labeled the key line jack RELAY and later amps have it labeled KEY IN.

The Alpha 89, 87A, 76A, 78, 77D have no adjustment for the full-on bias; it is set by a Zener diode in the various amp models. Later amps like the 9500 and 8406 allow for bias level change in software.

Paul, W9AC
User avatar
WA0VY
Posts: 83
Joined: Sun Apr 09, 2017 3:41 pm

Re: Alpha 89 vs. PureSignal

Postby WA0VY » Tue May 02, 2017 2:47 pm

For what it's worth, EBS in the OM amps can be turned off by a simple jumper on two pins on the control board. EBS in the Acom 2000A can be turned off by wiring a jumper in, which not hard to do but nothing like as simple as the OM's. I don't know about the rest of the Acom line. I can tell you that turning EBS off does make a noticeable difference on those amps.
73 Brent WA0VY
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Alpha 89 vs. PureSignal

Postby w-u-2-o » Tue May 02, 2017 3:31 pm

That's great info, Paul and Brent, thanks!

Should we change the name of this thread to something more generic, like "Electronic bias systems vs. PureSignal" or something like that?

73,

Scott
User avatar
VY2WW
Posts: 15
Joined: Mon Apr 10, 2017 6:31 pm
Location: PEI

Re: Alpha 89 vs. PureSignal

Postby VY2WW » Tue May 02, 2017 7:03 pm

Probably, Scott. Appreciate the comments. Will research the 89 further, but something is definitely going on to upset the PS calibration when the audio/mic input drops to zero.
WB2ZXJ
Posts: 5
Joined: Sun Apr 09, 2017 3:04 pm

Re: Electronic bias systems vs. PureSignal

Postby WB2ZXJ » Sun May 07, 2017 10:24 pm

.
Last edited by WB2ZXJ on Sun Nov 18, 2018 9:39 pm, edited 1 time in total.
User avatar
VY2WW
Posts: 15
Joined: Mon Apr 10, 2017 6:31 pm
Location: PEI

Re: Electronic bias systems vs. PureSignal

Postby VY2WW » Wed May 10, 2017 9:26 pm

Re: Alpha 89 causing Pure Signal to "chase its tail" under auto-attenuate, according to Alpha expert tech Bradley Fochen of KOHM Electronics, Colorado Springs (may be the last of the original ETO/Alpha team working on these), the 89 DOES indeed play games with bias, although it's not an EBS circuit, per se. The bias on the tubes is set very low re: the threshold of conduction/operation, so the curve is indeed hard for PS to "track." Apparently done to help "cooling" and tube efficiency. I'm debating the mod he recommends below, as I'm getting great results just avoiding auto-attenuate altogether and manually setting attenuation for the PS "green light," which, withe 10db added onto the coupler, just happens to be 24db on most bands. So all is well. Here's what he said:

"Try changing out D12 on the power supply to a 5.2volt zener. I must caution you that you might need to try other values to get the right results. The current value is 6.2v. You may need a value in between these. Do not go below the 5.2v and keep you talk time down. The tubes will get hotter at that bias.
Side note that the CW bias will also change when you do this."

His website is http://www.KOHM.net
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Electronic bias systems vs. PureSignal

Postby w-u-2-o » Wed May 10, 2017 11:14 pm

Very interesting, John, thank you!

73!

Scott
User avatar
w9ac
Posts: 290
Joined: Sun Apr 09, 2017 4:01 pm

Re: Electronic bias systems vs. PureSignal

Postby w9ac » Sun Jul 30, 2017 1:57 pm

"the 89 DOES indeed play games with bias..."

I don't understand what that means.

The Zener diode value change that Brad describes does nothing more than to adjust plate idling current into greater conduction over the RF cycle, moving the class of amplification further into AB and away from class B. The Alpha 89 does not use EBS, nor does it use any other form of dynamic bias management. Any games being played may have been done in production while optimizing a Zener diode value as the amp evolved over the years -- and indeed the 89 and 87A did evolve, particularly with extensive changes engineered into the PIN diode T/R system.

Amps like the Alpha 9500 and 8406 allow the user to accomplish the same bias change through software.

Paul, W9AC
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Electronic bias systems vs. PureSignal

Postby w-u-2-o » Sun Jul 30, 2017 6:38 pm

w9ac wrote:"the 89 DOES indeed play games with bias..."

I don't understand what that means.

Paul, W9AC

Paul,

John's (VY2WW) post of 10 May above explains everything. While not EBS, what the design does causes a dramatic memory effect that the PureSignal algorithm cannot keep up with, and it thereby enters a state whereby PS and the 89 fight each other:

- PS linearizes what it sees.
- The 89 reacts to PS by changing its bias state.
- PS linearizes to the new bias state.
- The 89 reacts to that PS change by going back to the original bias state.

This cycle repeats forever and causes auto-attenuate to fail, among other things.

73,

Scott
User avatar
w9ac
Posts: 290
Joined: Sun Apr 09, 2017 4:01 pm

Re: Electronic bias systems vs. PureSignal

Postby w9ac » Sun Jul 30, 2017 7:16 pm

>"John's (VY2WW) post of 10 May above explains everything. While not EBS, what the design does causes a dramatic memory effect that the PureSignal algorithm cannot keep up with, and it thereby enters a state whereby PS and the 89 fight each other..."

Sorry Scott, but merely saying that "...The bias on the tubes is set very low re: the threshold of conduction/operation, so the curve is indeed hard for PS to "track"'...does not explain the "playing around" with the Alpha 89 bias design. Using terms like "fight each other" to explain the cause doesn't help either.

Bias set at the verge of plate conduction is class B. Any amount of plate current conduction raises the class to AB. As reported in John's post, D12 does indeed affect bias, as does D13 (CW/SSB). However, fixed bias is set in the cathode, and is accomplished like most other grounded-grid amplifiers.

I don't disagree that PS performance can be affected by amplifier biasing and class of operation, but I don't see any reason why the 89 should be singled-out. Your original post indicated that PS sampling stability is affected with the Alpha 89 in-line -- but is it substantially worse than other grounded-grid amps with fixed bias? I see no such degradation to PS when using my Alpha 87A and 77Dx amps. My 87A's bias scheme is identical to that of the 89 model. My 77Dx is modified with EBS bypassed and is now biased exactly as the 87A and 89. I don't see PS hunting the RF sample.

If PS is noticeably degraded with the 89 when compared to other grounded-grid, fixed bias amps, I would like to see the fact-based measurements and causation.

Paul, W9AC
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Electronic bias systems vs. PureSignal

Postby w-u-2-o » Sun Jul 30, 2017 7:21 pm

Paul,

I don't have the schematic of the 89, and I don't have an 89 myself. I can only go by what I observed and what John reported. The fact remains that on his 89 there is more than a single bias state possible and that switching between those multiple states in response to PS linearization causes problems.

For a detailed breakdown of how the 89 works at the circuit level you are probably going to have to contact the same gentleman that John contacted, as he discussed above.

73,

Scott
User avatar
W2PA
Posts: 166
Joined: Sun Apr 09, 2017 6:34 pm
Location: LaGrangeville, NY
Contact:

Re: Electronic bias systems vs. PureSignal

Postby W2PA » Sun Jul 30, 2017 7:59 pm

There seems to be several versions of the Alpha 89 out there. Mine is a 1994-vintage one (with the ETO logo) and has no dynamic bias manipulation. You have two states from which to choose: SSB and CW, as someone's described earlier, which insert one or two zener diodes. PS has no problems with this version of the '89 in either setting.
73,
Chris, W2PA
User avatar
w9ac
Posts: 290
Joined: Sun Apr 09, 2017 4:01 pm

Re: Electronic bias systems vs. PureSignal

Postby w9ac » Mon Jul 31, 2017 6:27 pm

>"There seems to be several versions of the Alpha 89 out there. Mine is a 1994-vintage one (with the ETO logo) and has no dynamic bias manipulation..."

Chris,

Nice amp and I'm sure you enjoy using it. It's a shame that more of the 89 units weren't produced.

As you point out, the 89 and 87A went through a series of evolutionary changes, nearly all affect T/R PIN diode biasing. For example, later 89 and 87A units employ an additional Zener diode in the Rx PIN diode bias string that elevates bias an additional 150V-200V. In doing so, the Rx PIN diodes are better protected to ensure they do not conduct when line SWR is abnormally high, where voltage is nearly 1KV at full power into an infinite Z load.

A while back, Glenn Pladsen, AE0Q, at Alpha supplied me with a copy of the 89 schematic. If you or anyone else would like a copy, send me a PM request. The diagram shows no unusual biasing of the 3CX800A7 cathodes.

Incidentally, I recently obtained a NOS Alpha 86, still sealed in the original factory carton. Two problems though: (1) the power transformer was missing and its whereabouts unknown; and (2) it's a very early unit that's missing three circuit boards that are present in the final production units. By pure luck, a NOS transformer was available from K1XX who lives just south of me. We met up on I-95 for the exchange and I was set. If/when the PINs fail, I'll replace them with a pair of vacuum relays. K0HM mentioned above sells a nice drop-in vacuum relay board but he's not resuming production at this time.

The 86 went though several major production cycle changes. The amp I have uses low-voltage on the Rx PINs. Hard to believe, but the first units relied solely on a fast-acting VSWR trip circuit to protect the Tx and Rx PIN diodes. Moreover, Tx PIN diode bias current was insufficient and resulted in high IMD as shown in the QST Product Review. Two more changes later by ETO and they retired the 86 design while learning some valuable and expensive lessons. Those lessons resulted in the Tx PIN diode bias scheme used in the early 87A and 89,

Paul, W9AC
User avatar
W2PA
Posts: 166
Joined: Sun Apr 09, 2017 6:34 pm
Location: LaGrangeville, NY
Contact:

Re: Electronic bias systems vs. PureSignal

Postby W2PA » Tue Aug 01, 2017 1:56 am

Paul,

Thanks for the additional details and history. And the offer. I do have a complete set of schematics for the '89.

By the way I replaced the TX PIN diodes a few years ago. I sure enjoy completely silent QSK.
73,
Chris, W2PA
User avatar
w9ac
Posts: 290
Joined: Sun Apr 09, 2017 4:01 pm

Re: Electronic bias systems vs. PureSignal

Postby w9ac » Tue Aug 01, 2017 11:35 am

Chris,

If your plan is to keep the Alpha 89 for awhile, you may want to keep a small stock of Tx PIN diodes. The Microsemi/MACOM UM4006D is the correct OEM part number for the 4-40 threaded stud version. Richardson RF is closing them out at $24.75 ea. This is much less than original OEM pricing and far less than Alpha pricing.

Microsemi shows this as a discontinued, obsolete part. Quantity was 200+ a few weeks ago. 56 units are available as of today. I ordered a dozen to have on-hand for my Alpha 87A and some customized T/R switching projects.

http://www.richardsonrfpd.com/Pages/Pro ... tId=855450

Paul, W9AC
User avatar
W2PA
Posts: 166
Joined: Sun Apr 09, 2017 6:34 pm
Location: LaGrangeville, NY
Contact:

Re: Electronic bias systems vs. PureSignal

Postby W2PA » Tue Aug 01, 2017 9:26 pm

Thanks, Paul! I ordered two as backups.
73,
Chris, W2PA

Return to “PureSignal Operations (PowerSDR & Thetis)”