by msmpe » Sun Nov 01, 2009 11:14 pm
Here’s an update about my Steve Bench tube tester project. I finally ordered the parts I needed. I had all the PNs from Mouser for the SS stuff and was working on the resistors, when a friend gave me not only a bunch of used tubes (to test of course!) but a bunch of NOS carbon resistors. I would never use these for an amplifier, but seemed perfect for test gear. Good, saved me a couple bucks, but I had to sort through them all and organize them, so that ate up some time.
Then, many thanks to Tom McNally, I got a digital meter for the Gm read out. In the original SB circuit, Gm (and Mu) is read on an external DVM by measuring the drop across a 100 ohm resistor (R7). I’m installing a couple jacks for the DVM to verify, but the digi-meter is way cool! Now, one of these babies is not “plug-n-play.” This meter is set from the factory for + 1.999 Vdc. However, the SBTT is looking for a range of 10 to 130mVac across R7. The meter circuit can be changed internally by changing component values, but for the SBTT range, this involved some resistors, caps, and the calibration (I think) pot. An involved process fraught with risk!
The meter spec sheet also showed that an op-amp could be used to provide gain to the low level input. Seems simple enough, and it is, if you know how op-amps work! So I pulled out my old digital electronics book and restudied op-amps…
The first step was to determine the gain required. For low Gm signal tubes like the 12AX7, Gm is around 1,250 uMhos which should be a Vdrop across R7 of 12.5mV. At the other extreme, a high Gm power tube like the 6550/KT88 should exhibit a Gm around 11,500 uMhos which should be a Vdrop across R7 of 115mV. For the 12AX7 we want the meter to read 01.25 and for the 6550, 11.50. The gain factor for the op-amp is then Vout/Vdrop = 10. Sound right?
Gain for the op-amp is calculated: m=1+Rf/Rg where m is the gain factor, Rf is the feedback resistor, and Rg is the resistor to ground. Rf/Rg = 9, so any combination of resistors such as 18K and 2K will provide the correct gain. I decided that R7, Rf, and Rg should be precision resistors, that is .1%. Well for some reason, precision resistors come in odd values (E96?). Here’s what I ended up with: Rf = 8.87K and Rga = 976 and Rgb = 10, where Rga, b are in series. The error is .05% which is less than the .1% of the resistors themselves.
This digi-meter measures VDC, for the SBTT we need to measure VAC. My question: is all I need to do is put a cap in series with the meter to block the DC? or is it more involved?
Steve’s circuit can be broken down into about 7 "modules". The PSU uses a tripling rectifier off a pair of back to back 12.6vct trannys and the filter caps. My trannys will be external to the breadboard chassis.
Module #2 is a reference CCS with Q1 and a Zener string. I’m going to put the CCS with the small filter caps on one board.
The next module is the half wave negative cathode bias with Q6, VR1 (pot), and Zener string. Q6 is on a heat-sink. This will go on a board with another module that makes up the signal oscillator which uses a hex inverter IC and VR4 (trimmer).
The last perf board will carry both the plate and screen voltage control/regulator modules. The plate control consists of Q3, Q4, VR3, and R7. There is also a Gm/Mu switch which I am not including in the perf board set-up just yet. The screen control consists of Q5, Q2, and a triode/pentode mode switch. Q4 and Q5 are on heatsinks. If there’s room I’ll put the op-amp on this board. The digi-meter supplies the +5V for the op-amp bias.
The final module is the gang of sockets. I am going to only use one 9- and one 8-pin socket, parallel wired to a 9-terminal block and then manually switch leads from the modules to connect the pins appropriately. You could use the sockets and switches from a junked emission tester instead or use a series of sockets ala the SB circuit.
I have a layout done in AutoCad. I will post a picture when it is assembled (if I can figure out to include a jpeg). I’m using Plexiglas for the chassis with the boards mounted on standoffs above the chassis. Certainly not UL, but I’ll shrink tube all the leads, which will only leave the terminal block screws exposed.
I ordered my parts from Mouser for no reason other than I’ve never ordered from them before. Their project BOM capability is pretty cool, but you have to watch how you add to the project BOM, otherwise any added parts get mixed up with your cart and added to the project BOM twice! Should be simpler than that! Also they billed for sales tax...
Here is a list of the “critical” parts with Mouser PNs [formatting is description, symbol, qty, manfr, PN]:
74HC04 (not HCT) U1 1 ON 863-MC74HC04ANG
14 pin DIP connector for U1 DIP14 1 Tyco 571-1825093-3
MPSA92 350v PNP TO-92 Q1 1 ON 863-MPSA92G
MPSA42 350v NPN TO-92 Q2, Q3 2 ON 863-MPSA42RLRAG
IRF820 TO220 VFET Q4, Q5, Q6 3 Vishay 844-IRF820PBF
Heat sinks for the FETs HTSNK 3 Wakefield 567-690-3B
51V 5% .5 watt zeners CR9-14 6 Vishay 78-1N5262B
5.1V .5 watt 5% zener CR16 1 Vishay 78-1N5231B
10V 1W 5% zener 1N4740 CR17 1 Vishay 78-1N4740A
741 Opamp for meter U2 1 Texas Instruments 595-MC33078P
Dip connector for IC U2 - 8 pin DIP8 1 Tyco 571-1825093-2
100 ohm 1/4w .1% R7 1 Vishay 71-PTF65100R00BYBF
The total cost is just over $13. The biggest expense are the heat sinks, fortunately these can be recycled. The rest of the parts you can scrounge from the “JUNK” box, although you’ll also need a recycled PST, diodes, pots, sockets, jacks, etc.
Your comments are welcome.
8>) Mike
If there's no sound in a vacuum, where'd the music come from?