AO Gate Grinder

This is a module I’ve been thinking about since January. Nothing sexy, very much a utility module, but I was excited to get it together and working (pretty much) partly because it’s a more ambitious design project than any I’ve done before, and partly because it provides functionality I haven’t seen in any other DIY module.

Quoting the README at the Github repository:

Kosmo format synth module, 10 cm wide, to generate clocks and act on triggers and gates (called “signals” generally herein).

Signals come from three sources. There is a jack for an external CV input (±12V). There is a pushbutton for manually created gates. Finally, there is an internal clock to produce a stream of square gates. The clock can be switched On or Off and to Slow or Fast rate ranges, and its Rate can be set, on the front panel. (Slow is around 0.14 Hz to 15 Hz, fast is 3.2 Hz to 320 Hz using the specified components.)

If the Resize On/Off switch is set to Off, the OR of these three signal sources is sent to the Out jack. Otherwise the rising edges are used to trigger a one-shot whose output width is controlled by the Resize Narrow/Wide switch and the Width knob, and the resulting signals are sent to Out. Output widths can vary from about a few milliseconds to about 5 seconds.

(And yes, the width can be set to longer than the time between two pulses. It then ignores any pulses that come along until the resized pulse ends.)

The signal at the Out jack also goes to an inverter whose output goes to the ~Out jack. LEDs indicate when Out and ~Out are on. The rising and falling edges of the Out signal are used to generate trigger pulses which go to the On and Off jacks respectively.

So one can invert gates, change the lengths of gates, convert gates to triggers, convert triggers to gates, make a gate clock, make a trigger clock.

Repository at https://github.com/holmesrichards/GateGrinder includes KiCad format schematic, PCB design, and panel design.

7 Likes

It’s working, but there are some issues:

  • The first run of boards was too long. What was I thinking? About 15 mm of unused space at the bottom. It can be cut off without affecting functionality, although you lose the Analog Output logo :frowning_face: This has been fixed in the kicad_pcb file.
  • The output levels with the components I used are too low, about 3.9 volts — I’d intended more like 5V. I know where I went wrong. This can be fixed by changing the resistors in the output voltage dividers — though I’m thinking it’d be a good idea to replace them with trimmer pots. I haven’t made this change yet.
  • The trigger outputs fail to trigger some of my modules (they do trigger others), and I haven’t figured out yet if that’s because their amplitude is too small or their width is too short or what.
  • One very minor issue with the first run is the transistor footprint I used was the regular one and not the wide one. I like the wide footprint better generally, and specifically the BC547s I used have wide spaced leads. I got them in and soldered them without creating bridges, but I changed the footprints for any future boards.

Anyway, if anyone wants one of the first boards, let me know via personal message. I have two I don’t consider spoken for. I’ll request a few bucks to defray costs (negotiable), or barter is possible. I intend to do a PCB front panel but that’ll be a week or two probably.

6 Likes

Thats a pretty cool idea. Who says utility isnt sexy!

3 Likes

(So beggars, choosers, etc but is there a PDF or other render of the schematics?)

5 Likes

There is now:
https://github.com/holmesrichards/GateGrinder/blob/master/PCB/gategrinder_sch.pdf
https://github.com/holmesrichards/GateGrinder/blob/master/PanelPCB/gategrinder_panelPCB_sch.pdf

6 Likes

Excellent work, I may fire that through on my next JLC order.

I’ve had a short look at the schematic and I’d like to make a suggestion, if I may. You might consider replacing U2A (i.e. 1/6 of the 4069) by 1 transistor. And replacing the combination of U2B (another 1/6 of 4096) and Q2 by 1 PNP transistor (to make the logical opposite for Q4, obviously you need to rearrange the resistors currently connected to Q2 a bit as well). That would eliminate the 4096 altogether (you are currently using only 2 of its 6 ports, which seems a bit of a waste) and simplify the schematic.

3 Likes

Thanks, I did consider something like that. Decided at the time it seemed in some sense simpler to do it with the 4069 even if a bit extravagant with the ICs. I mean, they’re a mere 28 cents each at Tayda and it’s not like there isn’t real estate on the board… Though I hadn’t considered combining U2B+Q2 with a single transistor, maybe that changes the calculus…

3 Likes

Lies!

Looks very nice! What happens if you put lfo or audio thru it?

4 Likes

Probably triggers and very narrow gates at the input frequency? Haven’t tried that.

3 Likes

Still a couple of these available, I made a reverb listing:

5 Likes


Changed the 1.8k output resistors to 1.2k, now getting 5V outputs. Seen here with permanent front panel.

7 Likes

Clocking away:

Worked the first time even though I accidentally put in the wrong transistors and had to replace them, which the pads did not care for. I was going to go with a white button, but the shade clashed with the white of the panel. Yellow looks nice too. For LEDs, I stuck with the same color and used a faint ancient pair.

5 Likes

Just a quick question on the pads, I had to replace the precision reg on my Tuner VCO, and noticed the pads lifted on one. Having done repairs on comercial PCB’s in the past I have never had this problem but it seems a few people have had issues with PCB’s from JLC/WAY… Is this down to the Pad size or just that low volume production is less resillient?

Cheers

Rob

1 Like

Took me a while to figure out what the AO stood for in the title.

It’s funny because this is the first AO module I encounter and all its outputs are binary.

Looking at the schematic, with all the unused gates, I also felt like @Jos that it could be optimized.

You could connect the Manual Gate button to VCC with a pull down to ground and get rid of one of the inverters, you could then replace the two OR gates and the other inverter with a 4025 triple 3-input NOR device.

When you only need a couple of odd gates of different types it’s good to know about the 4572 which is a random mix of basic gates:
image
It’s old and relatively slow, but for our types of designs here, it should not be a problem.
You could also have used that instead of the two chips you used, but I think the triple 3-input gate would be a cleaner solution.

And why use half of a TL072 instead of a TL071? (Or a dedicated comparator IC for that matter, as you’re not using the other half as an opamp.)

You might also want to add a pull down on the OR gate input driven by the TL072, because when the TL072 is driving negative, that input is basically floating because of the blocking diode, and CMOS inputs don’t like to float. (No input does, but CMOS inputs really don’t like it.)

Just my 2¢ of, hopefully constructive, criticism.

3 Likes

Optimizing for price and availability is also a thing; e.g. TL072 is cheaper at Tayda, and they don’t have 4572 (Mouser has it, but it’s more expensive than a 4069 and 4071 combined). The fewer different components you use, the easier sourcing and BOM/stock management gets (cf. earlier discussions about trivial designs that require E96 resistors :smiley:).

(somewhat related, Dave Jones posted a video on BOM optimization recently, even if that mostly focuses on resistor & capacitor values and SMD machine constraints.)

5 Likes

Yeah, I mean, I don’t have any TL071s. I see them in circuits so far more rarely than TL072s, and it’s not like they’re any smaller and as @fredrik says, they’re a tad more expensive. Thanks for the other comments, good to learn this stuff.

3 Likes

Good question, I unsoldered a pot on one JLCPCB module (actually cut it off and unsoldered the pins) with no trouble but had three damaged pads unsoldering five resistors on another.

2 Likes

JLCPCB is crazy high volumes, but by default you get the cheapest option for everything, including the PCB substrate:

Their boards use low quality Tg130 or Tg140 substrate. It begins to burn and break down at normal soldering temperatures, so you have to be quick and careful or you’ll get lifted pads. We [OSHPark] use Tg180, which has a MUCH higher breakdown temperature. You can hold the iron on it as long as you want without damaging the board.

You get what you pay for, as usual.

3 Likes

Not just by default, JLCPCB says:

Typically, we use FR4-Standard Tg 130-140 for PCB production.

FR-4 Tg 155 has better flame retardance than standard FR-4, and it’s appropriate for lead-free assembly. We recommend choosing TG155 for the following conditions.

• 4&6 Layer PCB boards.

• PCBs use in automotive, military or aerospace applications

• High-precision PCBs with 4mil trace width/spacing, 15mil hole to hole clearance, BGAs.

but currently their order form doesn’t actually seem to offer anything other than Tg130–140.

PCBWay does offer higher grades as an option, but for a 100 x 101 mm board Tg130–140 is $27 (vs. $5 if it’s 1 mm smaller! And $4 at JLCPCB, $7 once you hit 103 mm.), Tg150–160 is $63, and Tg170–180 is $89.

For hobby stuff, at $82 cheaper I can live with the occasional lifted pad.

7 Likes