Eurorack DLF tempco and thermal glue

[waldenpotato]

If using a tempco resistor with the eurorack DLF, is it recommended to glue the resistor to the transistor pair? I don't see any mention of this in the build guide but have always tried to glue my tempco resistors.

[Synthbuilder]

If using a tempco resistor with the eurorack DLF, is it recommended to glue the resistor to the transistor pair?

I've never bothered with a filter. Indeed, I don't always use it for my own VCOs either.

It certainly won't do any harm and it will improve thermal transfer between the components. If you can do it neatly then go for it.

There is a good argument for putting a bit of insulation over the top of the PTC and transistors. This keeps the temperature even more stable and reduces the affects of air draughts on the pitch. Again, for a filter it's probably a little excessive, but for a VCO it might be worth doing.

Tony

[waldenpotato]

I was able to glue them together neatly without a mess. Attached is a photo.

[eggpie]

Been thinking about this a bit.... I had to buy ten of these due to the MOQ (LT7339002A1K0JTE) when I built the Flanger a while ago. How about carefully soldering a couple of leads to the chip PTC resistor and glue that to the ic. This could work for all manner of Tony's 5U and Euro vcos and filters, amongst countless other designs. The added benefits are cost and perhaps even better performance, as they are 3900ppm/K as opposed to the KRL Bantry 3500ppm/K.
A bit of a kludge / Heath Robinson I know. Any thoughts?

[Synthbuilder]

How about carefully soldering a couple of leads to the chip PTC resistor and glue that to the ic.

You can do this, and I have done so on a couple of occasions, and it'll probably work as well as any axial PTC resistor.

Having said that I've always had better results using a larger resistor and larger NPN pair. eg. THAT300 and KRL temp co. The larger thermal mass of the pair means that sudden changes in external temperature are less likely to cause a mismatch in temperatures between the two devices. Again, enclosing both devices in insulation will give a better response. One factor that is not often talked about is the self heating effect of the exponential transistor. As it passes current its temperature will rise which will lead to mismatch between the transistor and the compensating resistor. Using a larger geometry transistor should mean a lower temperature rise as the thermal mass of the device will take longer to heat up.

The added benefits are cost and perhaps even better performance, as they are 3900ppm/K as opposed to the KRL Bantry 3500ppm/K.

You actually want the overall compensation to be between +3300 and +3400ppm/K to match that of the Vbe junction in the exponential convertor. The +3500ppm/K KRL PTC resistors I use in the VCOs are used in a potential divider configuration so the resistor that sources current into the PTC will reduce the overall temp co to be around 3350ppm/K.

The actual temperature coefficient of the PTC resistor is often not well specified, so it could vary somewhat from the stated +3900ppm/K, with +3900ppm/K being the mean value of the parts sold. That said, I have tested several KRL PTC resistors from three different batches and years of manufacture and all tested to be between 3475ppm/K and 3525ppm/K so I think the KRL ones should be pretty accurate.

But PTC resistor compensation is not very precise as a rule. The actual temp co required for perfect compensation varies with temperature, and the self heating problem is ever present. I have found that even the KRL resistor with a THAT300 array doesn't work that well at colder temperatures, for example below 10C.

The axial through hole Arkaneohm PTC resistors are now no longer manufactured although world stocks seem to be quite high at the moment.