Mezmerize DCB1 Building Thread

[operatorius]

Would there be any noise issues if the toroid is close to other components within the chassis? Apparently I've bought a pretty narrow one so there are not many placement options.

 

[Salas]

Usually no, because insensitive X1 voltage gain circuit, but you can also wrap a single turn soldered at the seam copper tape around the transformer for magnetic shielding.

Place the transformer on the opposite side from the input relays in any case.

 

[operatorius]

Just finished soldering and mounting to the board. Here are the measurements:

Measured just right after the relay click:

Left out: ~7mV
Right out: ~2mV

Testing pad closer to IRPF240 ~ -9.6V
Closer to IRPF9240 ~9.45V

After a couple of minutes, when it warms up a bit,
Left out: ~4mV
Right out ~0mV (tester is pretty cheap but I've seen it go ~ -0.1 mV as well)

Testing pad closer to IRPF240 ~ -9.55V
Closer to IRPF9240 ~ 9.4V

I've read somewhere, that the negative supply should be lower than positive. could it be that I've overheated the mosfets? I've learnt the hard way to never ever solder large 3 pin components in place for testing purposes, removing them from the pcb is nearly impossible. Or at least don't bother mounting them to the chasis if you do not have lots of experience... this is my very first build, should have bought the radiators for mosfets.

 

[Salas]

What version board you got?

 

[operatorius]

What version board you got?

Ten years after

 

[Salas]

It basically works alright given your test numbers. If you want to try bring the rails closer replace with bit stronger VF led in the chain of five. IRFP9240 side.

 

[operatorius]

It basically works alright given your test numbers. If you want to try bring the rails closer replace with bit stronger VF led in the chain of five. IRFP9240 side.

Since this is my very first serious project, I have a goal to learn as much as possible. Spent some time measuring the components. They all come really close to each other. What I was able to discover though is that the LSK170 jfets are not B grade, but F! I was able to find just a single explanation for the F grade which means "free samples". I've measured their IDSS and discovered that they are on the lower end ~4-6mA. I am suspecting that the difference between +/- and the lower value comes from the fact the the jfets between bc550/bc560 measured ~5.7 and ~4.5mA. I have some with higher IDSS(but one is just bellow 6mA). Been digging the thread and saw you wrote that more close, higher idss ones should be placed on the leftmost position, bellow/above bc550/bc560. What about the rest? Also, should I just order a new b grade batch with higher idss from a more reliable source?

 

[Salas]

Higher Idss Jfet will drive Led chains total voltage drop bit higher. And vice versa. Those five Leds in each rail provide reference voltage. 4-6mA Idss samples is still a fine range to drive Leds, no problem with that. Just match those two Jfets driving the five Leds chains so to level the field for Leds. Given the Led chains are close in total voltage forward drop when wih same mA already.

Little different rails due to parts tolerances is affecting the DC offset just a bit in this circuit. Idss mismatch within pairs in the audio path quad Jfets is the main culprit for DC offset nonetheless. We have settled at a simple test standard for any DCB1 build i.e. the DC offset not to be above 5 mV on any of its two channels. We consider it very well made at this point. Also let it warm up a bit for measuring that.

AC performance is not affected within a logical DC results range. Audio will be as good as in a perfectly symmetric rails and zero offset example.

Striving for best DC results is still a noble learning path in parts selection and matching but don't ruin your build for that goal unless you got a vacuum desoldering station for safe and clean rework. Good solder wick and gel flux also help.