QUOTE: I am intrigued, resistors going low, not high?

Will look into this with one end of each R clipped out of circuit.

Re Resistors: Examine the DC circuit; Plate & grid resistors are normally OK if you measure across them. Bypassed cathode resistors with an electrolytic cap across it: No. B+ with filter caps in circuit: No.

If its high in circuit its likely a dud & any Mica, or Ceramic cap and resistor that happens to have an end free, here will be tested. One does not change caps without testing & replacing the dud resistors, as you go. To not do so is to risk re-work, & faults: Rework wastes time and increases the risk of damage.

Whilst Lafayette quoted below 50Meg in a decoupling cap and below 200Meg for a coupling cap, to be a fail. You will not find too many Wax Paper & Oil filled types that good, & a new 630VDC cap in a valve set, that tests infinity is a much better option. One needs an Insulation tester that can deliver a voltage at rated, or test value, to test them anyway (not for electrolytic type).

My valve & circuit tester actually uses a Neon powered with DC at close to common "B" voltage as an Non polarised cap tester: A cap should not pass DC. So ideally there is a blink as the cap charges & the neon extinguishes. If it does not, the magnitude of the glow reflects just how bad the cap is and you replace it.

Marc

Resistors going high or low....

A bit off topic but here are some generalizations, based on many years experience:

Carbon film or metal film resistors, when they do fail, and this is very rarely, always fail open circuit. How to identify? They generally are lacquer covered with metal end caps visible under the lacquer and/or rounded ends. This type of resistor is often used in more modern equipment as a fusible resistor, to restrict damage if a failure occurs.

Carbon composition resistor behaviour depends on the resistor maker and the value of the resistor.

IRC resistors (slightly rough phenolic surface, mid-brown body, square ends with rounded edges) tend to go high when stressed. Low values (say, less that 2k) tend to hold their value and eventually fail high.

Higher values (say 470k up) tend to go high for no obvious reason.

Ducon and Morganite resistors (smooth body, sharp end edges) less than about 2k will go LOW when stressed. Ducon particularly. Very bad if it's a cathode bias resistor. Ducon resistors are dark chocolate brown, Morganite, closer to milk chocolate brown.

Higher values tend to be more reliable than IRC. Of all composition resistors, Morganite tends to be the most reliable.

Of course, it's harder to tell the resistor type after it's obviously been cooked, but you should change such resistors anyway.

Half-watt IRC resistors 470k and over should be treated like paper capacitors.

To save unnecessary clipping of resistors when testing, read the circuit and do a little maths. You should never need to clip plate or screen dropper resistors. Also remember that 20% tolerance of resistor values was common in those days.

You might well clip a resistor, find that it measures OK, then destroy it when soldering it back into circuit.

"Think with your head, not your hands"! It's ALWAYS better to measure voltages first, rather than guessing.

Generalisations from those who have fixed a lot of radios, is the sort of observation lost in generation gaps & redundancies where information is lost as it is not written down.

I would further add that the most unstable area of a radio (other than a 6U7 without a shield) is the second detector 1st AF & coupling to a 6V6 & its family. I agree that 100K resistors & upwards are a common fail & 50K especially in Krieslers along with speaker transformers have an attrition rate.

For some reason the failure rate of plate resistors on 6B8 and most valves in that position is high as are the grid one resistors on 6V6 and others. I do not stock many 1/2 Watt resistors due to voltage ratings & inventory. To get closer to correct resistance in AGC you can often use 1/4 Watt. as some sets used 1.75Meg in a divider & 1.8Meg you can get.

Marc

BringBackTheValve:

QUOTE: I am intrigued, resistors going low, not high?

Oops, I just glanced at schematic and see those four 100k's are in series-parallel. It seems nine-times-out-of-ten one can get away with measuring resistors in-circuit in valve circuitry, but not with this anode-follower arrangement. Since amp is now at an associate's workbench, went there today and only one 100k R was, in fact, out of spec (measured high outside of 10% tolerance - 113k)(replaced.)

Ian Robertson:

QUOTE: ..RFCs..restrict capacitive loading on the amp due to..capacitive reactance of electrostatic speakers

Yes, appears to be the case as amp previously bench tested with only full range cone speaker connected. Amp goes berserk, it turns out, if electrostats are not connected! Once everything is connected, the .0033 cap was then successfully reconnected in feedback with continued stability.

But when I suggested "do the math(s)" I mean it.

When I'm checking resistors in, say, a TV chassis, I have a spreadsheet open that contains the calculations for a number of network combinations. I just enter the resistor values.

What I then do is S/C the power supply (set not powered of course) with a clip lead to convert the network into (usually) a Thevenin equivalent circuit. Then I measure the resistance of the node I'm interested in to ground. The spreadsheet tells me what I should be seeing.

If that checks out but I'm still unsure, I pull the valve if necessary, power the chassis up and measure the voltage to see if that matches the value calculated in the spreadsheet.

This is all DC circuit theory 1.01.

Much quicker, safer and neater than clipping and soldering and it avoids unnecessary disruption.

After a while you'll find you can do the more basic calculations in your head.

Golden rule of fault-finding - disturb as little as possible.

I'm looking at circuit of QUAD mk II amplifier (EF86's driving KT66's) and see it also uses anode-follower-inverter with strange local feedback to first cathode (do not understand how that works) that is added to global feedback (with no compensation capacitor!) It also has more local feedback from output transformer to KT66 cathodes! This amplifier would be a good candidate for analysis with LTSpice, I can obtain transformer inductance measurements to 'plug into' LTSpice as a friend has a pair of these amps.

Clearly, Philco was influenced by British post-war hi-fi radicalism, including Peter Walker's research papers on Electrostatic speakers resulting in the QUAD ESL-57 "Walker's Wonder" in 1957, just a year or two before Philco's imitations!

QUOTE: Load the .ASC file from my post.

That link returns a 404 error

I had a problem with the copy I downloaded from the site.
When I extracted it from the email I sent to Brad, that copy works fine.
So if you un-hide your email address or email me I'll send it to you.

Here is sound of this amplifier finally running, with original Electrostatic panels: https://www.youtube.com/watch?v=3mtU_iK4fsc

One method used to prevent ultrasonic oscillations in tube audio amp circuits is to use "grid stoppers". These are resistors, mounted one end right at the tube socket control grid pin, that, combined with the tube's Miller capacitance form a low pass circuit to pass audio gain but reduce any gain at ultrasonic frequencies. Thus preventing parasitic oscillations. Usual values are around 1K for output tubes, and 68K for a 12AX7.

As pointed out; Never leave the OP tube grid resistor/s unchecked, even the grid stoppers fail, that frequently, that I consider it a rate of attrition.

Good tips there about grid resistors/stoppers - I think this amp needs them as many products were rushed into production back in that frenetic boom era of consumer electronics.

The most common grid 1 resistors for 6V6 250V in older sets was 500K (grid leak) to ground, 50K (grid stopper) to grid, and naturally, input into the join of the two resistors.

One of the instabilities I have found is when there is an RF stage, or RF feeding into the audio amp. Whilst I consider it can have the advantage of sustaining the OP valve at a more constant current. Some valves I have seen in audio (6M5) (42) (6V6) (6A6) (807) are not real fussy & are quite happy to amplify it. and that's when it becomes a real problem.

Replacing paper "B" filter caps with Electrolytic can also not help and you need to add an NP cap of around 0.1mfd to decouple. The electrolytic is not a good decoupler of RF.

A lot of the stray RF here rides on the mains & gets in that way, which is why I resort to filters on the mains to try and get rid of as much of it as I can.

QUOTE: sound of this amplifier finally running, : https://www.youtube.com/watch?v=3mtU_iK4fsc

This is funny:
I clicked on the 'Show More' on this You-Tube upload and it turns out their (google) computer identified the music that was playing! - it was sourced from a classical FM station -

Google A.I. supercomputer says it's "III. Allegro con spirito" by Chamber Orchestra of Europe