Jonny Tubetronics: amps and synth stuff now in one thread

Brill 👌 wafer switches and discrete components, the proper old school way of doing things, chuck in a relay or two and your all the way.

Sssshhhh sounds great, look forward to seeing the end result.

^ Excellent - cheers.

I've run out of time to progress this any further this week, but the good news is these jack leads turned up. This means I can connect some things up properly next week and make some funny noises. Looks like I've got enough resistors to start the hokey noter. Getting excited now!

I breadboarded the hokey noter. Idea being I can stick some voltages into the VCO without committing to soldering anything permanently.





And I'm glad I did, because it didn't do what I thought it would do.

I started with a 5V regulator and made up a potential divider of 5 lots of 120k Ohm resistors in series. That gives 6 voltage points of 0V, 1V, 2V, 3V, 4V and 5V. Should be one octave between them, so between 1V and 2V, the 120k resistor was made up of a chain of 12 10k resistors in series to make up the 12 semitones in an octave. Ran out of space on the breadboard, of course, so had to loop back on myself a bit to fit it on. I then stuck a jack lead in the 1V/Oct socket and on the breadboard connected it 0V with a flying lead to touch on any of the voltages. They weren't really notes and I didn't really get 1V per octave; it was closer to 2V per octave if my ears are right and felt a bit more logarithmic than linear. And the 0V point didn't play the bottom octave on the VCA, it was one octave up, so I re-arranged the breadboard to put the 12 10k resistors between 0V and 1V. Measuring the resistors, they're not going to play accurate notes, but they should at least play something close that sounds out of tune. So either I haven't understood how 1V per octave works or I've badly tuned the VCO or I've soldered something up wrong. At this point I'm not sure and some further investigation is needed.

Good stuff! It’s definitely keeping my brain working!

I thought I'd resurrect this since I'm updating my valve amp and this will potentially allow me a second chance to kill myself with high voltages. I sort-off stopped because of spiralling costs and nothing much to show for it. I'm a little over £300 in at this point, which only really includes the LMNC modules, and I need to add quite a bit more before I have anything useful to play with. The only bit that 'works' so far is the LMNC VCO and that doesn't really do much by itself because it (ideally) needs a voltage source of some description to make it generate notes and then filters if I want to make interesting noises.

First I ought to finish the valve oscillator, since I started buying bits for it. Since I put down my initial ideas for this, I realised that square and sawtooth waves are more useful for a synth than sine waves. I also learned about generating those on a valve circuit by tapping into different points of the timing system. That's covered in the Wireless World articles in the this post, so I'm planning to build the circuit as-is but I'll allow space to tap into the timing circuit and allow space to stick in another valve if I need it. This could end up doubling-up as a valve function generator.

So, I've bought a thing to progress the project: a power supply:



If it looks familiar you'd be right, because it's exactly the same as the one I used in my pre-amp. The only difference is this one still has its smoothing choke. No idea what it's off, but it's from the same eBay parts robber seller and I really liked the compact, integrated form. It makes it easy to work with and almost what you might call modular, and it fits nicely in the metal Hammond enclosures I like to use.

Next thing I bought a transformer for the VCF. You can see it in the top right of the circuit on the output of the EF86 and it forms part of a feedback loop:



Its function isn't completely clear. I would have thought you could use a resistor and decoupling capacitor here, but the author has stated: I could have used the plate of the pentode directly as the feedback signal, but I included a transformer, to provide a low impedance. (Its nothing special, I used a 230V to 24V 2W power transformer. The other option would have been another tube. The 200k at V2s plate is used to provide damping to the tuned circuit which the transformers inductance forms with the parasitic capacitances. Adjust it to your needs.)

So I've gone with what they did and I can fiddle with it later. I found a little 24V 2W on eBay which should do the trick:



Next up is to try and decide what type of LDRs are needed. I'm really not familiar with using them so it might be a bit of trial and error to get them working. I'm also wondering if I can swap out the voltage summing op-amp at the bottom of the circuit with an ECC83 driving a couple of those luminophore lamps I've got (or even the MTH-90 lamps or IN-3 lamps I tested at the bottom of this post). That'll be something for later though - one thing at a time...

The PSU arrived and is indeed identical to the other one, bar the choke. That meant all I needed to do was quickly check all the wires went to the same places, check everything was grounded to the chassis, no shorts, no broken wires, then I was happy to attach some mains and test it. First without the rectifier valve to check for sparks, smoke and explosions, then with the valve and repeat. Here it is up and running:



Healthy 290V DC on the HT and the Mullard EZ80 rectifier valve glowing away nicely:



The other transformery thing that I thought was a choke isn't a choke, because it's got two windings. It's not connected to anything on the PSU. One winding goes to two wires on the B9A header and the other is connected to the RCA jack on the end panel labelled EX LS:


Can't decide what that means. EX maybe means External? LS maybe Line Speaker, Loudspeaker, Line Source? The outer winding to the B9A header measures 535 Ohms DC resistance and has a 5.8k resistor and 0.001uF cap in parallel with it. The inner winding to the RCA jack measures 0.4 Ohms. I think that rules out the RCA jack as an audio input, because the impedance is too low.

I put 5V rms on the 535 Ohm winding at various frequencies which resulted in the following voltages on the RCA winding:

106mV at 50hz
109mV at 100hz
110mV at 200hz
111mV at 500hz
111mV at 1khz
110mV at 2khz
106mV at 5khz
94mV at 10khz
66mV at 20khz



So it's about 50:1 ratio. The fact that it keeps going up to about 10khz means it could be a low quality audio transformer, but the input impedance seems a bit low for a valve output transformer. 100V line speaker maybe? Answers on a postcard to the usual address.

For the valve oscillator I've been looking at the design parameters in the Wireless World article to adapt it to my needs. The example they worked through was for a fixed frequency of 50hz with a 22V output. I want a variable frequency and ideally I'd like to use the big variable capacitor I've got so I can sweep through a range of frequencies. The timing circuit is the R1, C1 and R2, C2 pairs on the far left:



R1 and R2 are the same value, and C1 and C2 are the same value. My variable capacitor is a 2x 525pF so it will slot in where C1 and C2 go and therefore give me a variable frequency. So that it's not just a random range of frequencies, I can calculate a value for R1 and R2 to set the lower frequency and add two more capacitors across C1 and C2 to limit the upper frequency. If I don't add the additional capacitors, the upper end of the frequencies will be extremely high and heavily squashed up at the top end of the scale. To make all this easier, I put the equations into an Excel sheet so I could easily fiddle with the values:



I aimed for a 20-200hz range to start with (to see if it worked) and that showed I needed 13.7M resistors for the bottom end and 56pF capacitors to limit the top end. To then get higher frequencies, all I need to do is switch resistors R1 and R2 by a factor of 10 smaller to go up a decade, so 1.37M for 200-2000hz, 137K for 2000-20000hz etc. That'll be done with resistors on a wafer switch to switch in the new values. I can alter the range to do 10-100, 100-1000, 1000-10000 by doubling the resistor values etc etc. That bit is quite simple.

The next bit I wanted to do was to try and adapt the circuit to use 2x EF86 pentodes instead of an EF86 for V1 and an EF91 for V2. This simply would have meant all the holes I drill are the same size, all the valves I buy are the same, all the valve bases are the same, same size spacers and so on. For V1, they have done all the hard work already because it's already worked out for an EF86, but I followed the process through using an EF86 datasheet just to be sure I was doing it correctly. I got slightly different values, but only different enough that I put it down to maybe some slight differences in the datasheet they may have been looking at compared to the one I've got.

The thermistor is something which doesn't have a lot of information covered online. I think it's there because the circuit starts to resonate from zero and the amplitude basically increases to a point where it needs regulating. The thermistor is where this happens and it alters the voltage in the feedback circuit to regulate the amplitude. My numbers called for a thermistor value of 4.4k with a dissipated power of about 42mW; a 4.7k 50mW thermistor seems to be a common part so I'm 80% (ie mostly) confident that this part of the circuit will work with a bit of tweaking. I think this means that either R4 or Rth will need a trim pot in series.

Where I started to trip up was selecting resistor values for V2. There's a little inconsistency between the two worked examples which I need to go back through the article and check, but either way the values required lay outside the resistor values in the tables on the datasheet. This itself isn't a problem because the valve doesn't have to operate between strict limits, so I very roughly extrapolated the values on Excel graphs to extend the values for Ia, Ra (R6) and Rk:





However, because the relationship between the gain and feedback from V2 back into the circuit is quite critical, this is where I lost my confidence in my selected values and I'll need to re-read the article before I'm happy an EF86 will work here. I think I may need to build this circuit as-is, initially, and work on changing the valves and output voltage later. So off the back of that, I've ordered an EF91 to use for V2 (which wasn't expensive) and I can use the Wireless World values for now. The EF91 is another miniature valve with a B7G base, but the key difference between the EF91 and the 6AS6 is that the EF91 can take 250V whereas the 6AS6 can only take 180V. As I mentioned before, 180V is a bit more challenging to get down to from 300V than 250V, so the EF91 is preferable to the 6AS6 from that perspective.

The LDRs arrived so I had a go at seeing what they did with an old-style green LED. I figured I’d want to check a couple of these so I soldered the LED to a length of wire with a 330R resistor on one leg and heat-shrinked the legs to make it easy to handle:



Then I could tape the LDR facing it and measure the resistance at different brightnesseses from 0-12V:



For this one I got about 1.5M when dark and about 500R when bright, so quite a decent swing. If I were to put this in the feedback loop of an amplifier, for example, it would effectively become a voltage controlled amplifier (VCA).

I started marking up the case for the oscillator. The challenge with this thing is going to be fitting everything in, because the variable capacitor takes up a lot of space.