Maybe I need to get out more, but somehow this mess (see my earlier post) annoyed me. So I decided to round up some valves and see if I could make some once-and-for-all definitive measurements. Here's what I did and what I found.
METHOD
I tested the valves individually, one after another. I connected just the heater to a variable DC supply (Coutant, labelled 6V, 5A but with no obvious serial number, adjustable between 4.5V and 7.9V). I measured the current and voltage into the valve using two DVMs (nothing special, Mastech MY64s which I bought from Argos a few years back). I arranged the circuit so that I could read the voltage with a precision of 0.01V and the current with a precision of 0.2mA.
I started each test with the psu at 4.5V and over the first minute I adjusted the heater voltage to 6.30V if I was testing E-series valves, or adjusted the current to 300mA if I was checking P-series ones. Then I left the valve running for a further 4 minutes, tweaking the supply to keep the voltage at 6.30V or the current at 300mA as the valve warmed up. There was a very significant increase of heater resistance during the first minute. After 5 minutes it would still be rising, but very slowly - less than 0.2% per minute (all the valves I tested were the same in this respect). So I took the measurements at the 5 minute point and those are the results I've listed below.
I've already mentioned the measurement precision. I have no way of knowing what the absolute accuracy is. But suffice it to say the two Mastech meters agreed with one another and also with a third DVM from a different manufacturer to better than 0.5%. The results were also very reproducible. I picked one valve (one of the Mullard PCC88s) to be a reference and I measured it regularly throughout the test sequence. Out of a total of 14 measurements of this valve's heater voltage, measured over 21 days, I got 7.17V 6 times, 7.18V 5 times and 7.19V 3 times. I call that pleasingly reproducible
.
I tested a total of 61 valves - 22 PCC88s, 21 ECC88s and 18 E88CCs (or equivalents). All of the valves were vintage, not modern. The E-series ones come from my collection. Most are used and unboxed. The exceptions are five Brimar E88CCs and one Brimar ECC88 which appear to be NIB (new, in box). I didn't have any PCC88s at the start of this exercise, so I had to buy some. All but three of these were described as NIB by their sellers and, looking at them, I think that only one of the NIB ones was actually a used pull, put back into the box from which its replacement had presumably been taken.
RESULTS
5 Mullard ECC88s, coded GA2 B***, 332mA, 367mA, 378mA, 384mA, 395mA
1 Mullard ECC88, coded GAG triangle***, 376mA
5 Amperex ECC88s, coded GA* triangle***, 358mA, 367mA, 376mA, 378mA, 397mA
2 Pinnacle ECC88s, coded 660, 371mA, 375mA
1 Telefunken ECC88, coded ?9102704, 333mA
6 Brimar ECC88s, uncoded, 304mA, 308mA, 310mA, 315mA, 317mA, 340mA
1 Brimar ECC88, coded 4B1/1778, 291mA
6 Brimar E88CCs, uncoded, 300mA, 301mA, 302mA, 302mA, 303mA, 306mA
2 CV2492s, Brimar coded 3L2/1778 and 3J1/3910, 292mA, 296mA
5 Mullard E88CCs, coded 7L0 R**, 305mA, 306mA, 307mA, 310mA, 311mA
1 Mullard E88CC/01, coded 7L1 R***, 301mA
4 CV2493s, Philips/Mullard coded 7L1 R***, 295mA, 295mA, 303mA, 304mA
5 Mullard PCC88s, coded DJ2 B***, 7.17V, 7.33V, 7.39V, 7.45V, 7.58V
5 Mullard PCC88s, coded DJ3 B***, 7.15V, 7.20V, 7.25V, 7.47V, 7.60V
1 unidentified PCC88, coded DJ3 symbol**, 7.05V
1 Tungsram PCC88, coded triangle***, 7.19V
1 Valvo PCC88, coded triangle***, 7.04V
2 Tesla PCC88, coded 517 A4 37, 6.68V, 6.81V
1 Ultron SQ PCC88, coded 11203, 6.68V
3 Mazda PCC88s, coded 8MO (with another manufacturer's markings crudely rubbed off), 6.56V, 6.62V, 6.63V
2 Mazda PCC88s, uncoded (but same boxes and valve labelling as the above 3), 6.00V, 6.50V
1 Mazda PCC88, uncoded (different labelling from the other Mazdas), 5.79V
CONCLUSIONS
So, what have we learned ?
ECC88s have a wide range of heater currents. They seem to fall into two groups. The ones from the Philips companies (Mullard, Amperex), and perhaps the Pinnacle ones too, fall into a group with an average current of 373mA and a wide range - from 332mA to 397mA. The Brimars average 312mA and range from 291mA to 340mA. The Telefunken could fit into either group. The way I read it, Morgan Jones was saying that it's the Philips E88CCs which have the 365mA heaters. But my measurements show that it's their ECC88s which do. The Brimar ECC88s are more like 300mA (Jones measured 10 of them and he found this too).
The E88CCs are much, much more tightly controlled. The Philips ones (Mullards and CV2493s) average 304mA and range from just 295mA to 311mA. This is consistent with my Mullard data sheet which says they should be 300mA. Indeed they are. But so are the Brimars (including the CV2492s). They average 300mA, ranging from 292mA to 306mA.
The PCC88s can be split into three groups. The Philips coded ones (Mullards, Tungsram, Valvo and the one whose ID I couldn't read) run at an average of 7.30V, ranging from 7.04V to 7.60V. Perhaps Philips changed the spec between 7V and 7.6V during this valve's lifetime ? This would be a neat explanation for the confusion in the data sheet, for the average value and for the wide range. But frankly they're not distributed in two tight groups around 7.0V and 7.6V. They look pretty evenly scattered to me. So
The next group consists of the "8MO" Mazdas (which were clearly made by someone else), the Teslas and the Ultron. These range, between them, from 6.56V to 6.81V with an average of 6.67V.
That just leaves the three other Mazdas. One of them, at 6.50V, might belong in the second group but the other two, at 5.79V and 6.00V just look like rebranded E-series valves to me. Whether this was careless or deliberate I don't know.
The bottom line is that if you want a "proper" PCC88 with a 7V+ heater you'd better buy a Philips one. If you buy a Mazda you 'll get something with a lower heater voltage and if you're unlucky you might actually get an ECC88.
If you buy a Philips ECC88 you'll get a valve with the same heater power as their PCC88s. If you buy one from someone else it will likely have a heater which is closer to a "standard" E-series valve (6.3V, 300mA).
If you buy an E88CC, you will get a valve with a
very precisely controlled 6.3V, 300mA heater and that seems to apply both to the Philips ones and to the Brimars I tested.
If you use the "wrong" Philips (i.e. a PCC88 in a 6.3V system, or an ECC88 in a 300mA one) then its heater will be under-powered. Of course that could have been what the circuit designer wanted.
So I hope that's put this one to bed
Cheers,
GJ