Voltage regulators and input impedance


I have an Audio Research Ref 5SE pre running into a Modwright KWA 100SE power amp. The input impedance of the Modwright is listed as “15K at 50 Hz” and the specs on the Ref 5 suggest “20K ohms minimum load”. 
 

So here is my question: I have gone through 3 6550 tubes in the Ref 5 (which AR says is used as a “voltage regulator” in the power supply) in the past 3 years - sovtek, tung sol, svetlana, all new. The last one lasted under 500 hours. I am thinking of potential causes, including excessive heat (the location needs better ventilation), something wrong with the preamp, etc, but I am wondering about that input impedance on the power amp. Am I running the Ref 5 into too little of a load? Would that put undue strain on the power supply tube, causing it to fail prematurely? Would love to hear from someone who knows this stuff better than I…

128x128ilikemiles

Why don't you fire off an email to Audio Research - their customer service is exemplary.

Also check with Modwright - it may be very easy for a local tech to increase the input impedance of your power amp to 20k without impacting the amp. If this is possible the overall sound may be substantially improved through better matching.

A loading impedance of 15K will not damage the preamp or its tubes. All it will do is impact the lower frequencies and you will not get optimal performance from the preamp. My guess is that AR uses a low value coupling cap on the output for sonic reasons and is using the 20K minimum input impedance of the connected amp to make up for it.

 

Tubes going bad could be a symptom of high heater voltage. Check your wall voltage. If it's greater than 120V it could be a problem for tube gear, especially if their transformer specs are based on a 115 volt primary.

If the 6550 is acting as parti of a voltage regulator circuit in the PS, the slight impedance mismatch you cite would not stress it. More importantly, what is the input Z of the Modwright at 1kHz? There’s not much music at 50 Hz. Historically, AR were known for driving power tubes hard, near their limits, and power tube failure was a common problem, but I don’t know if that’s still true, and anyway the 6550 that’s failing is not a power tube, if I understand you correctly.

Excess heat will shorten he life of any electronics, including tubes. I easily get 2,000 hours on the 6550 in my Ref 5SE, and the tubes are still working when I replace it out of caution.

At either 15kOhms or 20 kOhms your current is extremely small. Take a 1Vrms signal your current is only 0.05 or 0.07 milliamps respectively.

Trust me, your amp’s input impedance is not causing your tubes to burn out because of maybe 20 microamps more load.  It is much more likely however to soften the high frequency response a little.

Perhaps there’s a part failure or a bad solder joint in the VR circuit that results in excessive plate dissipation ( too much wattage) across the 6550. You need to have it examined by a competent tech or by ARC directly.

My guess is the impedance issue is a red herring so far as the VR tube failures are concerned. I think you wrote above that ARC specifies an input impedance of not less than 20 K ohms. This suggests that the output impedance of the pre-amplifier is around 200 ohms. (I didn’t look it up.) An actual 15 K input impedance, even in the low bass frequencies, is not going to over stress the preamplifier in any way. And anyway the input Z of the Modwright rises to 23K ohms at 1 kHz which sneaks in over the lower limit.

As I understand it, the only risk of going below that 20K amplifier input impedance is that the bass may roll off prematurely.

I do not believe it's related to your power tube issue.

Well, it’s not that the bass will roll off, although I guess it would at a barely measurable rate since the midrange input Z is stated to be 23K ohms, as that once you get below a ratio of 1:10 for output Z vs input Z, you start to lose more signal voltage to ground, which could eventually, as the ratio gets progressively lower, result in noticeable loss of gain. But 1:10 does not represent a wall or cut-off point, the fraction of signal energy transferred from one component to the next is a continuum on that ratio. 1:10 usually preserves more than 90% of the signal energy. It goes down from there as the two impedances become more and more equal. At 1:1 ratio, 50% of signal energy would be lost to ground. In the case of these two components, I doubt that the problem with the Voltage Regulator tube has anything to do with the impedance match, which is actually acceptable, even at 50Hz.

Definitely contact Audio Research… last time I had a question I called. They were very helpful. I don’t remember the tech guys name now.

I owned a REF 5SE… I changed one tube after 2,000 hours only because it was recommended. I upgraded to the REF 6SE before any other tubes needed changing.

Definitly something is wrong.

Well, it’s not that the bass will roll off, although I guess it would at a barely measurable rate since the midrange input Z is stated to be 23K ohms, as that once you get below a ratio of 1:10 for output Z vs input Z, you start to lose more signal voltage to ground, which could eventually, as the ratio gets progressively lower, result in noticeable loss of gain.

That is absolutely correct - when I distributed tube gear in the 80’s we provided a modification service to customers power amps to increase the input impedance to provide an optimum load to match the tube preamps - a significant benefit in terms of "apparent gain" and overall sound quality.

I just want to correct what I wrote on 06-01.  I guesstimated the output impedance of the PH5SE at 200 ohms, based on the recommendation to mate it with an amplifier of not less than 20K ohms input Z.  In fact, if one applies the 1:10 rule of thumb, I should have guessed the output Z at 2000 ohms, not 200 ohms.  In fact, ARC say its output Z is 300 ohms unbalanced/600 ohms balanced. Mea culpa. But in view of the facts, I wonder why ARC recommend to drive an amp with not less than 20K ohms input Z.  In theory, the PH5SE could work just fine with an input Z much lower than 20K ohms. Could be that the output Z of the PH5SE rises at frequencies other than 1 kHz, where the measurement is usually done.  Also, input capacitance presents an impedance that is in parallel with the resistance.  This can cause a high frequency roll-off if the input/output impedance ratio is not a healthy greater than 1:10 ratio.  Anyway, the main point is that the matching of the two components is almost certainly not causing the 6550s in the ARC PS to blow or wear out prematurely.

Thanks for the sharing of considerable expertise here. I will contact ARC and see what they say.