Strange input loading effect.. or not?

[exclamationmark]

I ordered up some other SSR's, namely the CEL 7141E series, Omron G3VM-351B and some extremely pricey Panasonic RF relays. All of them pretty much floored the Vishay LH1525 in bleed-through comparisons, with the CEL 7141E's having a little less than half the capacitance, around 20-30pF. The capacitance appeared to be pretty linear as well. The Omron relays were almost the same as the CEL's, but with a slightly higher capacitance (and only 350v rating as opposed to 400v on the CEL's). The panasonic RF relays pretty much did as expected, with sub 10pF capacitances (I honestly didn't know if I was measuring noise or an actual signal at lower frequencies!) . A trend I found in all the devices was that the capacitance often decreased by about 30-50% as I swept the frequency from 500Hz to 10kHz. I don't think frequency dependant capacitances would cause any signal distortions, as you can essentially model this by sticking a whole bunch of capacitors and resistors in a ladder type network, with however many poles you wanted (plus 20-30pF shouldn't even be audible). The thing that WILL cause distortion is the bias voltage vs output capacitance - a similar thing happens in X5R and X7R ceramics. Though since the capacitance only gets lower with increasing voltage (and any effects should already be inaudible), I still don't anticipate any problems.

I haven't tested any of these new relays for their resistive linearity when turned on, but the Vishay LH1525's I originally used tested to be almost perfectly linear from 50uA to 20mA, with a resistance of roughly 25.3 ohm per device. I'll test my new SSR's sooner or later, but I don't anticipate any major differences in linearity, except at higher currents.

To me the CEL 7141E's are the clear winner, as their capacitance isn't too bad and they're dirt cheap! The only thing upsetting about these is the fact that the 6 pin DIP version (only a single switch) is the same size as the 8 DIP double switch version. If anyone hasn't caught on yet, I'm in the midst of building a programmable tube preamp. I'll actually have to re-design my board to accommodate this fact, as everything is literally surface mounted (I use 6 pin sockets with bent leads so they can be surface mounted), and space is extremely tight (perhaps an understatement). I could use the 8 DIP version of this relay, but I don't really feel comfortable with gap of only 1mm between pins and potentially 300 volts between them. Oh well. If everything goes to plan perhaps you'll see this contraption in the builds section of the forum.

[Steve Conner]

Those devices from CEL come in surface mount packages too, why don't you use those if you're worried about space? You can cover both sides of the board in the suckers and make Mesa jealous.

Another possibility might be to make a standard little board with a bunch of SSRs, resistors and an I2C expander chip like the PCF8574 to drive them. This would be your high voltage digital pot or rheostat module, and you could construct and test them in bulk, and mount a load of them vertically into a motherboard along with your tubes and stuff.

[exclamationmark]

Those devices from CEL come in surface mount packages too, why don't you use those if you're worried about space? You can cover both sides of the board in the suckers and make Mesa jealous.

Another possibility might be to make a standard little board with a bunch of SSRs, resistors and an I2C expander chip like the PCF8574 to drive them. This would be your high voltage digital pot or rheostat module, and you could construct and test them in bulk, and mount a load of them vertically into a motherboard along with your tubes and stuff.

I was tempted to use the tiny smd relays, but I looked at my board routing and it won't actually save me any space, except in a couple places (but overall my board dimensions will still be the same). I think it's really the size of my other components and the minimum trace spacing I'd like to use that's limiting my board space. I'm using these teeny tiny 3 watt SMD resistors for plate resistors but there's NO WAY these can dissipate 3 watts without burning something, hence being oversized (I think the spec sheet specifies a delta T of like 160 degrees C at 3 watts), and 2512 size surface mount resistors for everything else. I'm even using smd film capacitors. I'm not entirely sure how these will sound though, so I might have to rip them out later and tack on some 'real' capacitors.

I actually tried that vertical mounting module thing a while ago with reed relays and it worked pretty well. I had them sandwiched in between two PCB's, to keep the logic and HV sides separated. It looked kind of like the old tube cordwood modules you used to find in 'compact' equipment of the day. The only thing was that my modules were like 4 inches high so it forever shattered my dream of mounting it in a 2U rack unit. Here's a pic of one of the modules I ripped out of it. There's supposed to be another row of header pins on the top but I must have de soldered them or something. The IC on the top is just a bog standard shift register, IIRC I controlled something like 80 relays using only 3 I/O lines!

[Alan0354]

Finally, when I removed the relay across the grid stopper, the scope trace at the output suddenly shot up to be on par with the reference circuit. Tracing my way back to the grid of each triode, I found the voltage to be much lower on the side where the relay was attached. Again, when the relay was removed, the trace went up to meet the reference circuit.

What voltage are you referring to, please specify where you measure and what is the voltage with and without the relay. Also do you mean the relay is across the 470K resistor R4 or R5? Capacitance across these two resistor will not roll off hf.

[exclamationmark]

What voltage are you referring to, please specify where you measure and what is the voltage with and without the relay. Also do you mean the relay is across the 470K resistor R4 or R5? Capacitance across these two resistor will not roll off hf.

At 5kHz the voltages were about 1.25V and 1.13V respectively (first measurement is without the relay, the second is with the relay), measured at the grid with the same triode. The relays were across either R4 or R5, as it did the same thing in both circuits! A 65p capacitor across either of these resistors mimicked the behaviour of the SSR as well. I have a feeling it has something to do with the actual way a tube works though (like reverse leakage or grid current or something. I don't have any suitably large resistors to test it right now, but I'm sure if you replace the grid with a 10Meg+ resistor to ground (to mimic the grid resistance), the capacitive shunt would have no effect, apart from the intended treble boost.

[Alan0354]

I spent a little time reading the data sheet and even went on Vishay site to look for application notes. There are problems with this relay when use for 470K. Fig. 12 is the isolation and is specified at 50 ohm termination. If you look at Fig.12, the isolation is -40db at 1MHz. For 50 ohm termination, this mean the reactance from S to S' is 100X50=5000 ohm. You back calculate using C=1/2\piRf and put in f=1MHz, you get 31pF from S to S'. For input of the tube where impedance is way into mega ohm, you don't have any isolation. That's the reason you see much sharper clipping with the relay in. signal just go from S to S' whether the relay is on or off.

You might have problem with the SSR for R8, R9, R10, R11 for the reason as above. You might not see it as your are only shorting out one of the 470K, you still have the other one. And also the plate resistance is only 10K, low enough to mask the problem. Bottom line, I just don't think these relays are meant for working on such high impedance environment.

Try the other brand, read the data sheet, be aware of their test condition and have to kind of translate to your working environment ( 470K resistor) and see whether it will work. Problem is even if the SSR is for RF, they are for 50 ohm system. It might produce problem for 470K. 50pF is not a lot for 50 ohm and will not cause problem, but it'll drown a 470K. You might want to consider real DIP relays.

BTW, Why are you putting a relay across the grid stop resistor? You know you need one already. 470K for grid stop is very high. 100K can soften the blow and will not affect the sound already.

Regarding to roll off, I can only see it roll off at higher frequency when the relay is off and if the output capacitance is from S' to LED. Depending on the output capacitance, if you calculate 470K and 30pF, the -3db frequency is 11.3KHz. So you definitely have roll off when you insert the relay across the 470K resistor if the output capacitance is from S' to LED. Whether you short one end of the LED to ground is irrelevant because your driver is low impedance and is ground referenced. I don't see any data regarding to capacitance change with frequency. You should not have any roll off when the relay is on as it short out the 470K already and you claimed your driving impedance is low.

[Alan0354]

I still have no idea how the shunt capacitance was decreasing the frequency response, but I did find one other tidbit of information concerning the rather large capacitance. Apparently MOS capacitors are frequency dependant, and most manufacturers state their capacitance values at 1MHz, probably in an effort to look impressive (the capacitance is lowest at high frequencies). The problem is, at audio frequencies the capacitance increases due to some mumbo jumbo about minority carrier mobility. So 30pF on a datasheet might turn out to be 100pF at audio frequencies!

I am not aware of MOS capacitance is frequency dependent. I only know capacitance is voltage depend as the depletion layer get thicker when voltage is higher.

[Steve Conner]

I'm totally puzzled by the report of signal level decreasing when the relay is put in circuit. Seems to me it should do the opposite. :|

[Alan0354]

I'm totally puzzled by the report of signal level decreasing when the relay is put in circuit. Seems to me it should do the opposite. :|

I don't either, that's the reason I said the only thing possible is if it happens only when the SSR is off, then the output capacitance is 30pF from S' to LED ( grounded either from direct connection or through the output of a driver with low value resistor like 3K). But I doubted there is 30pF between the S' to LED as it is opto coupled device and the fact it can standoff high voltage implies there is a good physical separation between the LED and the MOSFETs. I think that's the spec for 0.75pF. The problem is the data sheet did not specify most of the condition and explain how the capacitance are measured.

It dose not make sense, that's why I did spent the time going through the data sheet and try to find the app notes.