some questions about design choices on the Skoolie

[Dimitree]

hi everyone,
for fun and as a learning tool, I was comparing the schematic of the Skoolie with the Total Recall / Tourbus, as I find interesting to learn how to convert a design from positive ground to negative ground.
I noticed some differences and I'd like to know why these were needed (if they were needed for the positive to negative conversion of just to improve the design)

1) there is a missing resistor (68K) between pin 5 and 6 of the NE570, and the other two resistors (R9/R10) have a lower value (47K instead of 68K). Also, on the second half of the NE570, the two resistors R17/R18 have higher value compared to the others schematic.



2) those two capacitors were added on the circuit before the BBDs



3) for the second BBD, the pin 8 is tied to the VG point of the first BBD (and so, dependant on the BIAS1 trimpot). On the other schematics, the pin 8 is tied to the voltage set by the BIAS2 trimpot (pin 5 of IC10).



4) on the Skoolie the voltage feeding the opamps is the VC point right after the 7815 regulator, and the voltage for the LFO/BBD is taken after a 22R resistor. On the other schematic, it is the opposite, the LFO/BBD is taken right after the regulator, and the audio path is supplied by the "isolated" voltage.



many thanks and sorry for the dumb question! :-)

[thomasha]

1) there is a missing resistor (68K) between pin 5 and 6 of the NE570, and the other two resistors (R9/R10) have a lower value (47K instead of 68K). Also, on the second half of the NE570, the two resistors R17/R18 have higher value compared to the others schematic.

I read somewhere that the described changes were done because of the supply voltage, and not polarity. MN3005 chips work nice at 18V, MN3205 only work up to 9V. The MN3005 also works at 9v, where there is lower headroom. Some resistors were changed to reduce the gain of the NE570, so that it will not hit the BBD as hard. The missing resistor is likely also related to this change. If you check the DirtBag discussions there was also an additional capacitor to filter some signal and improve the sound when using the MN3205.

I guess changes in 2) are a result of changes in 1). If you increase/decrease the compression you also have to increase/decrease the expansion.

3) is more complicated, since Vgg depends on the positive/negative ground approach. With -15V it should be -14V (considering the datasheet).

[madbean]

You've made some excellent observations here. Every one of those changes was done deliberately. I'm not at my work computer ATM but when I get the chance I'll explain the reasoning behind each.

[Dimitree]

thank you really much, that would be awesome and really interesting!

[Dimitree]

just upping this post, in case you forgot :-)

[madbean]

Yes, I did but here are my notes.

1) The Skoolie is based on the later negative ground adpations of the Memory Man, namely the Stereo Memory Man (which is the mbp Prognosticator project) and the Echo 600. In those circuits, the 68k is omitted. You can include the 68k, but what I found is that you end up having to use a large value DC coupling cap otherwise you get a bunch of crap in the signal. And, it doesn't improve or change the circuit in any positive way. In fact, I partially traced the EHX Memory Boy many years ago and they did use a 1uf ceramic with a 68k in series. But after messing around with so many variations of the circuit in the ensuing years I never really found any advatage to including that vs. eliminating it altogether.

R9 and R10 are lower values (47k vs 68k) because of what happens in the second half of the COMPANDER chip (the expandor portion).

On the second half of the compander, again this was based on the Stereo MM. Pins 10, 11, and 12 have a nominal output of 1.8vDC by design due to the internal supply voltage used in the chip and the two internal resistors it has (20k/30k voltage divider). The chip is designed to modify the voltage outputs of thos pins by giving the designer an option of an extneral pulldown resistor on pin12. So, when you follow the math it uses, the value of R17 ends up bumping the output voltage upwards by some amount. The idea here is to pick a value which results in a DC output level that is close to the virtual bias point (IOW the mid point voltage) of the circuit. So, by bumping that output from 1.8v up to around 7 or so, we get much higher headroom on our delay output.

2) In the DMM, those caps are tied to ground. For a negative ground adpation, there are two choices: you can tie those resistors directly to +VC or use a DC blocking cap and tie them to ground. Again, I went with the option closest to what EHX did in the Stereo Memory Man.

3) This part is not obvious but I'll explain: the VGG voltage for both BBDs in not dependent on the setting of either BIAS trimpot. If you look more closely, you'll see that the BIAS1 trim and R50 form a FIXED voltage divider. It doesn't matter where the BIAS1 trim is set because the junction of BIAS1 and R50 is basically two fixed reisstors whose values result in about 1vDC, which is exactly what the VGG pin of the BBD requires for a negative ground circuit run @ 15v. IIRC, I used the same voltage divider for both the VGG pins on both BBDs b/c it made the layout easier. The VGG pin on BBD2 could just as easily been tied to the BIAS2/R55 junction point and I would have done so if the layout would not have been a mess.

4) This weas purely a design choice on my part. IMO, it makes more sense to have a small current limiting resistor on the BBD to ensure that the +15v supply limit on that chip is not exceed rather than the other way around. The important part here is that even if the supply rail with the 15v regulator ends up being slightly on the high side, you are ensured that the MN3005 stays a little under that (as opposed to the other ICs the circuit uses which are not limited to 15v max).

[Dimitree]

thank you, everything is clear! 

I also spotted another pair of differences, on the Expander section, what is the advantage in using C17 and C18, instead of a single cap (the original uses a single 4.7uF cap)? Same for the Compressor section (added C8, 2.2uF).

[madbean]

thank you, everything is clear! 

I also spotted another pair of differences, on the Expander section, what is the advantage in using C17 and C18, instead of a single cap (the original uses a single 4.7uF cap)? Same for the Compressor section (added C8, 2.2uF).

I read a guide a while back on compander use and it suggested using separate decoupling caps on each of those pins rather than straight connecting them. I think the claim was it could result in possible noise reduction. EHX did this on a couple versions of the Memory Man, I believe.