The quick answer is that you can interchange the 6DJ8 with the 6922 in any equipment. The 7DJ8 has a slightly higher voltage and lower current filament and was intended for series filament television sets. It's close enough to 6.3 V and will just run a little cooler, the heater a little darker (toward red) and will have a bit more noise, but that is more a matter of selection than design. Because my designs use series filaments it is important for all tubes to draw the same filament current and therefore should be of the same maker. I was measuring some 6DJ8 filaments the other day and found the following range. Tungsram .294 amps, EI .382 amps, 6922 USSR .315. The spec is .365 so the EI is the closest. I suspect the Tungsram is actually a 7DJ8 marked 6DJ8. This kind of stuff happens all the time in the tube industry. The 7 volt tube was more popular in European TV's and they probably had a lot left when solid state took over. They simply labeled them 6DJ8 because we would want them over here. For all I know, the factory was already closed and we were just buying existing inventory. Like those horrible 6C33c tubes the Russians are dumping over here. That's not even close to a good audio tube. It was designed as a series pass tube in MIG jet power supplies. Bias stability, forget it. There is always some tube that nobody wants that someone hope to find (or make) a market.
Following is what I think is important about small signal tubes like 6DJ8/6922, 12AX7/7025 and the like. I always tell people to ignore what is printed on the glass, that can be anything. You have to find out what is going on inside and that's what the rest of this story is about.
At RAM TUBES noise testing is the main event. It's very difficult to do and I think we are the only people doing it properly or at all. In 1978 I started looking into the noise characteristics of 6DJ8's for the Beveridge RM-1 preamp. At that time only Mike Moffat's Theta and my RM-1 were using those tubes. I made a manual tester that could handle a couple of dozen tubes a day. I was also curious how noise changed as the tubes aged. Sometimes I left a batch in for a week, taking data several times a day like a doctor with a patient in the critical care ward. With the yield of phono input tubes being less than 10% I knew I had to develop some way to test a lot more tubes efficiently, gather the data, sort and match from that data. So in 1981 I built what was the first computer driven tube tester. It took most of a year to work everything out so I could test 32 tubes per batch. Once I had this going I bought bulk boxes of 100 tubes from various manufactures and saw some very interesting things. Sylvania/Phillips couldn't make a quiet 6DJ8 at all. GE wasn't much better. RCA sold European tubes with their brand on them and switched suppliers whenever they wanted to. The PQ Amperex did not last very long and the Russians were not yet available. The best thing going were the Tungsram, still in production in Hungary. I settled on those and 12 went into every RM-1. They were selected so the quiet ones were where they needed to be (phono input, Post RIAA amp, and line input).The not so quiet ones went where they didn't cause any problems (cathode followers and intermediate gain stages. By doing this we could use 95% of the tubes. You don't need a quiet tube everywhere and that is why RAM Tubes come in 4 grades. Tungsram branded their tubes 6DJ8/6922. In the industry thats called a "double brand" and implies that the tube meets the specs of both numbers which may or may not be the same. I was sad when they closed that factory in the early 80's. We switched over to the Russian 6DJ8's and generally got good results but, now and then, we got a batch that was either noisy, low gain or horribly microphonic. I remember one time having to borrow some money from my dad to buy 18,000 6DJ8's from a particularly good batch.
The 4 digit number system was intended to designate industrial tubes which generally were either made to or tested to a higher standard then their commercial counterparts. Thus a 6922 is just another number (and may or may not be better) than a 6DJ8. When the numbering system was set up by RETMA (Radio Electronics Television Manufacturers Association) there were agreed upon specs. Nowadays, all that has gone to hell, so when Sovtek creates some new number variant there is no one to object. That's why the list of 12AX7's had grown in number and is virtually out of control. All those letters that follow the tube number like LP and LPS etc are just made up Sovtet. RETMA would have put a stop to that immediately. You wouldn't see GE, RCA or Sylvania doing things like that. Back then is was hard enough to get an "A" after a tube. You really had to have a good reason and get approval.
I plan to write up how the system is set up at some later date. Of course if I get lots of encouragement it will come sooner. There are lots of little quirky things in the system, but it is a system at least. Somethings I have read, some I have been told and some I have figured out by myself. I assure you there will be some surprises in there.
Nowadays, some tube vendor might take any 6DJ8 and put 6922 on it. He might take a 6922 and put 6DJ8 on it. He might test the 6922 to a higher standard or not at all. Tektronix selected their tubes for their scopes for parameters that were important to them (gain matching, low bias drift, etc) and put silver labels and serial numbers on them and charged a lot for them. There is no universal test for any tube and the best tests are the ones devised by circuit designers (like me) who know what conditions the tube will be operated in and how it must behave.
So to your question, from a functional point of view there is no difference between a 6DJ8 and a 6922 and one is just as good as the other. Audiophiles who think 6922's are better are living in the past when there was extra attention paid to those tubes, but not always for audio reasons. I've spent most of this week tuning up the RAM small signal tube tester. I can measure output noise on a 6DJ8 down to 6 Microvolts. To compare it to a signal lets take a 0.6 mV moving coil cartridge. Since the cartridge is connected to the input of the tube we have to refer our 6922 output noise back to the grid by dividing by the tube gain which is 30. That makes the input noise of the tube 0.2 microvolts and the signal to noise a healthy .2 uV/.6 mV. which works out to 70 dB. That''s about as good as you need on a record player because that's around the S/N of the record you are playing.
Just think about that for a moment. That is 0.2 millionths of a volt of input noise. I assure you, it's not easy to measure due to the demands it puts on the test setup to be low in noise also. In essence, my Apple II driven tube tester has all the low noise requirements of an RM-4 headamp.
One more example. If you look in an RCA tube manual you will find a tube called 7025 and see that it has just two specs. One for one for noise (hiss) and one for hum. They are 1.8 uV of grid noise and 7 uV of hum. Hiss wise thats 19 dB more than a good 6922/6DJ8. Hum is very dependent on the circuit layout and we all know DC is necessary if you are making even a MM phono stage. The book refers you to the 12AX7A for all other specs. RCA mostly likely met this spec by selection rather than special design or construction. Every tube was tested anyway (one at a time by a real, usually female, person) and all they had to do was add those two tests to the regular battery of tests.
Turns out the noise of a good 12AX7 is not 19 dB worse than a 6DJ8 but when both tubes are selected to be as good as it gets there is about a 10 dB noise difference. Thats why I use 6DJ8's in all my low noise applications. Every tube, transistor, FET has a theoretical noise limit that one can not get underneath.. that's why it's a limit.
This practice is still the norm for many electronic parts, transistors, IC's, rectifiers and zener diodes. It explains why you can buy 100 volt 1 amp diodes (1N4000 series) and find that almost all of them are good to 1000 volts but not always and not to be depended upon.They try to make them all good to 1000V then they test each one. They know the relative ratios of how they sell and fill the bins accordingly. Once the high voltage bins are filled they will continue to throw 1000 volt diodes in the 100 volt bin. If you note, the whole 1N4000 series has very little (if any) price spread because they all come out the same now. When the system was set up for this diode that was not the case and high voltage ones did not occur on a regular basis and therefore, sold at a premium.
