This is what I use and I hope this sheds a bit of light on what we look for in a good valve.
THE “AVO” VALVE CHARACTERISTIC METER Mk. IV
The problem of designing a Valve Testing instrument capable of giving a true and comprehensive picture of the state of any valve has always been one of considerable magnitude, increasing in complexity as new valve types are brought into general use.
For a quick general purpose test necessitating a minimum of time and technical effort a mutual conductance figure will give an adequate idea of a valve's usefulness and the original “AVO” Valve Tester as designed to test the efficiency of valves on this basis.
Whilst a Valve Tester must, of necessity, be accompanied by a data hook correlating the results of the Tester with the condition of the valve in question a purely empirical figure if used as a standard, will always give rise to doubts in the mind of the operator. The instrument should therefore, produce a figure, which can be compared with some standard quoted by the valve manufacturer if the operator is to use his instrument with confidence. For this reason the “AVO” Valve Tester used the static zero bias mutual conductance figure as a basis or comparison, this figure being at that time almost universally quoted by the valve manufacturer.
In order to reproduce this standard correctly, it was also necessary to reproduce the stated values of DC anode and screen voltage a matter of some considerable difficulty when it is realised that for any stated condition of anode and, or screen volts die corresponding electrode currents can vary over very wide limits, and in the case of valves of initial anode current and high slope, the actuation of the control which produces the milliamp-per-volt reading might easily double the anode current flowing. With DC methods of testing the inherent internal resistance of the rectifying circuits used could be such as to give regulation errors which could cause results to be meaningless unless complicated thermionic stabilising circuits and a vast army of monitoring meters were used in all voltage supply circuits. Such complications would not only render the Tester of prohibitive price and size, but would considerably increase the complication of operation for the non-technical user.
The problem was overcome by the introduction of the AC method of operation (Patent No. 480752) by which means the necessary DC test conditions were correctly simulated and a true mutual conductance figure produced by the application of AC voltages of suitable amplitude to all electrodes. This enormously simplified the power supply problem rendered regulation errors negligible, and obviated the necessity for voltage circuit monitoring. The “AVO” Valve Tester thus fulfilled normal testing needs for a long period.
THE “AVO” VALVE CHARACTERISTIC METER Mk. IV
The problem of designing a Valve Testing instrument capable of giving a true and comprehensive picture of the state of any valve has always been one of considerable magnitude, increasing in complexity as new valve types are brought into general use.
For a quick general purpose test necessitating a minimum of time and technical effort a mutual conductance figure will give an adequate idea of a valve's usefulness and the original “AVO” Valve Tester as designed to test the efficiency of valves on this basis.
Whilst a Valve Tester must, of necessity, be accompanied by a data hook correlating the results of the Tester with the condition of the valve in question a purely empirical figure if used as a standard, will always give rise to doubts in the mind of the operator. The instrument should therefore, produce a figure, which can be compared with some standard quoted by the valve manufacturer if the operator is to use his instrument with confidence. For this reason the “AVO” Valve Tester used the static zero bias mutual conductance figure as a basis or comparison, this figure being at that time almost universally quoted by the valve manufacturer.
In order to reproduce this standard correctly, it was also necessary to reproduce the stated values of DC anode and screen voltage a matter of some considerable difficulty when it is realised that for any stated condition of anode and, or screen volts die corresponding electrode currents can vary over very wide limits, and in the case of valves of initial anode current and high slope, the actuation of the control which produces the milliamp-per-volt reading might easily double the anode current flowing. With DC methods of testing the inherent internal resistance of the rectifying circuits used could be such as to give regulation errors which could cause results to be meaningless unless complicated thermionic stabilising circuits and a vast army of monitoring meters were used in all voltage supply circuits. Such complications would not only render the Tester of prohibitive price and size, but would considerably increase the complication of operation for the non-technical user.
The problem was overcome by the introduction of the AC method of operation (Patent No. 480752) by which means the necessary DC test conditions were correctly simulated and a true mutual conductance figure produced by the application of AC voltages of suitable amplitude to all electrodes. This enormously simplified the power supply problem rendered regulation errors negligible, and obviated the necessity for voltage circuit monitoring. The “AVO” Valve Tester thus fulfilled normal testing needs for a long period.
Dont buy used tubes, new tubes already have enough probs.
For audio thay are useless, they will last about 200/300 hours.
At VTS used ones will last about 500 hours, but Andy will inform you about each tube in question.
Used tubes are tipically sell as strong(senile) on Ebay.
Last edited:
Only buy used tubes for curve tracing purposes or if dirt cheap. NG for audio typically.
The gm of a tube drops considerably while very new until reaching a steady value for most of it's lifetime, then gm drops rapidly near the end. So a gm measurement will tell you almost nothing about the remaining lifetime. Some testers had a "life test" button which reduced the heater voltage to see if the gm was about to start dropping. Useful if all compared tubes are tested on the same tester, but non standard for comparing tubes between different testers.
The zero bias gm measurement likely gives a similar hint, since that would tend to be checking for maximum gm develop-able. (the "life test" reduced heater voltage button has a similar effect by bringing maximum develop-able gm down into the normal tube test range.)
You can also look for burn (glass darkening) marks on the glass tube, especially around holes in the plate or off the end of the cathode. Fairly subjective. I wouldn't buy an obviously browned out tube.
The gm of a tube drops considerably while very new until reaching a steady value for most of it's lifetime, then gm drops rapidly near the end. So a gm measurement will tell you almost nothing about the remaining lifetime. Some testers had a "life test" button which reduced the heater voltage to see if the gm was about to start dropping. Useful if all compared tubes are tested on the same tester, but non standard for comparing tubes between different testers.
The zero bias gm measurement likely gives a similar hint, since that would tend to be checking for maximum gm develop-able. (the "life test" reduced heater voltage button has a similar effect by bringing maximum develop-able gm down into the normal tube test range.)
You can also look for burn (glass darkening) marks on the glass tube, especially around holes in the plate or off the end of the cathode. Fairly subjective. I wouldn't buy an obviously browned out tube.
Last edited:
gm is transconductance, usually measured in mA/V. mmo (actually, mmho) is millimhos, which is another way of saying mA/V. The two are equivalent: one is the parameter, the other is a unit with which the parameter is measured.RhythMick said:
Look for boxes which look suspiciously pristine - may be fake.
I have bought hundreds if not more used tubes and i guess 80 to 90% of them are very good. I go into the sale expecting the worse and that way when the tube turns out to be very good sounding that is a plus. Signal tubes tend to be better overall, than power tubes which are more of a crap shoot. Hi power tubes like el34, 6550, seem to have less life left in them then say el84 or 6l6. If you are going to pay serious money for a power tube that tube should come with some kind of guarantee.
If someone quotes a number with no units and no context then you can be reasonably sure he doesn't know what he is doing. It is likely that the numbers are either current or transconductance but with the anode voltage and grid voltage unspecified they don't tell you very much. At the very least they should say what device they used to do the measurement. You may also be suspicious of measurements which are all very close to the datasheet value.
I disagree. I have many used tubes, some as old as the late 1920s, that still perform well in an amplifier. Type 45 and 80 seem to last a long time, as do 56s.
TV-7 speak.
The ubiquitous mil surplus TV-7 type testers give a direct numeric readout like 72 or 64 with a specified minimum. If you use a TV-7 type, this is adequate information. There is a chart to convert that numeric to transconductance.
The Redneck Boatanchor site had a lot of information on mil spec tube testers. Here is a mirror to a small part of the original site with a file for conversion of the TV-7 readings to mmho:
Nolan Lee's Redneck Boatanchor Page
e.g. a random NOS Sylvania 45 that I grabbed from a shelf here at the office read 89 on my TV-7/D. That converts to 2225 micromhos.
Tubes with similar gm are not necessarily matched; so when an eBay seller says they are matched, they may or may not ( most likely ) be.
Win W5JAG
Or... They may know exactly what they are doing, hoping that the buyer does not...
I don't know anything about other testers, but the TV-7D/U types seem pretty conservative on their minimum spec. I do not toss a tube that reads at or below the min on a TV-7 until I have tested it in circuit.
For example, I have some globe 45's that test below minimum on my TV-7; they work fine in my Tubelab SE and modified Tubelab SSE. As do some 417A's.
On the other hand, tubes can test quite well on the tester, but may or may not work that great in a particular circuit. At least that's my limited experience.
So I wouldn't get that wound up over, or put a lot of reliance on tester readings. You really need to put them in the circuit you are using - that's the test that matters.
Win W5JAG
For example, I have some globe 45's that test below minimum on my TV-7; they work fine in my Tubelab SE and modified Tubelab SSE. As do some 417A's.
On the other hand, tubes can test quite well on the tester, but may or may not work that great in a particular circuit. At least that's my limited experience.
So I wouldn't get that wound up over, or put a lot of reliance on tester readings. You really need to put them in the circuit you are using - that's the test that matters.
Win W5JAG
I have had fairly good luck with used tubes as well to echo rkoonce's comment. Some high hour tubes like vintage MO KT88 are a pretty good risk IMVLE. I have old globe 45s that were used and served well for a decade in an amplifier that I gave away recently. They slumber, awaiting the day when they shall rise again in a new, and as yet to be designed and constructed amplifier.
I use an AVO 160 which has the very useful ability of being able to test most tubes under the exact conditions listed in the data sheet, or the conditions of use if different. I also have a Hickok 539B which is now languishing having been supplanted by the AVO. I also have a uTracer which is very useful for evaluating different (often unfamiliar) tube types for operating characteristics and linearity, it is also useful for generating spice models.
I use an AVO 160 which has the very useful ability of being able to test most tubes under the exact conditions listed in the data sheet, or the conditions of use if different. I also have a Hickok 539B which is now languishing having been supplanted by the AVO. I also have a uTracer which is very useful for evaluating different (often unfamiliar) tube types for operating characteristics and linearity, it is also useful for generating spice models.
sharks... do not dramatize that much. I have had like 90% luck with evil-Bay purchases, the other 10% not THAT good were kinda mentioned as "used" in any form which I might have overlooked, but that's really not a big deal as been lab-ratted while the initial bread-boarding or in further metal works on chassis, no problem.
I collected maybe 25 used 45's 15 to 20 years ago when you could still get used ones cheap at hamfests and on EBAY. In fact all of mine were purchased for under $15 USD each, most under $10. Some were sold as "weak." All of them work fine in an SE amp (a Tubelab TSE) where they operate on 325 volts at 30 mA.
I found an old RCA data sheet that listed a set of operating conditions for AB2 push pull on 275 volts resulting in 18 watts output. I figured that I should get over 20 watts on 300+ volts, so I put together a test amp. Only a few 45's out of my collection were up to the task in that amp. Why?
In the TSE the tubes idle at 30 mA and may see peaks near 60 mA when played loud. In the push pull AB2 amp the tubes idled at about 15 mA each, but were called on to eat peaks of about 200 mA. The 70+ year old filaments had plenty of emission to deal with 60 mA peaks, but fell short when asked for 200 mA. There were 5 or 6 tubes that worked well and I saw over 20 watts on 325 volts into a 3300 ohm OPT, but I didn't have enough tubes that were good enough to justify building that amp, so I went with NOS 307A's.
This experiment proves that the usefulness of a used tube depends on what you intend to do with it.
I had 3 different tube testers two were mutual conductance (Gm) testers. In nearly 10 years of using them I found very little correlation between tube tester readings and usefulness of a tube as an audio amp. This was especially true for output tubes in amps where high peak currents were drawn, but also for small signal tubes with regards to distortion.
Years ago I designed a power amp for HiFi and guitar amp duty. The amp used a 6U8 for the driver / PI and KT88's for output, and ran at a conservative (for KT88's) 35 WPC on 425 volts, due to the early vintage Chinese tubes propensity for blowing up, sometimes in a spectacular fashion.
I purchased a bulk package of NOS RCA 6U8's from a reputable supplier, and ran them through the tube testers finding 4 bad tubes which were replaced by the tube seller. After the amp design was finalized I found that the distortion numbers at a medium power output level varied a lot as I rolled all 100 tubes through the amp, with 3 or 4 tubes classified as "gross outliers" reading over 5% THD while most were well under 1%. Being an engineer in my full time job, I computed the mean and sigma for the lot , and there were a few (don't remember exactly how many) that were outside 3 sigma, and the 3 or 4 that were really bad (outside 6 sigma).
Experiments proved that the outliers were really sensitive to screen voltage on the pentode section for distortion while tubes near the mean were not sensitive at all. A tube tester will not find this!
I built 3 of Pete Millett's Engineers Amplifiers. They use 6CB6's in LTP for the input stage. I had to roll a large box of used tubes through the first amp to find two pair that had low distortion and good DC balance. NOS tubes were better, but some tube rolling using a dual trace scope and distortion analyzer was needed to find good ones that matched. Again tube tester readings were of no help.
After a few more similar finds, I sold my tube testers!
I now have a vacuum tube curve tracer, but have yet to really run through a large box full of identical tubes.
I still buy used tubes at hamfests. I will buy random unboxed used sweep tubes for $1 each or less, boxed ones up to $3 if they look OK. I avoid the common audio tubes in used condition unless there is some kind of guarantee.
Buying small signal tubes that are usually cheap, in used condition usually doesn't make sense. Tubes like the 417 / 5842 which have become expensive may be a good deal used. I have about 50 that were "pulled from military spares" nearly 50 years ago. They have been quite variable in the TSE, so I plan to see if the curve tracer is useful for predicting usability in an amp.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.