Great to see the regular posts on here @phosphenes 
I’d happily pay extra for a 4-later PCB, especially if you prototyped a 2-layer & saw a noticeable difference, even if only a slight one.
I’d suggest using the standard 4 op smt op-amps used in most non-TH (through-hole) designs, I’ve forgotten the name but it’s similar to ADA4851-4. Presoldered would be lovely, & again, I’d pay extra for that.
The LM6172 has become to expensive in my opinion.
1 Like
I’m definitely intrigued to see the difference between a 2 layer and 4 layer on a DIY build, probably a worthwhile experiment for the DIY community in general. I do wonder if the benefits of the layer stack up would be insignificant compared to the issues introduced by through-hole components etc.
I’m going to stick designing with the LM6172 for the time-being because its what I’m used to but I do take your point, I’d love to be in a position to offer pre-soldered SMD but that feels a bit further down the road for me!
1 Like
I would definitely go for a 4 layer PCB, whether it’s thru-hole or not. Just design a 2-layer board like you normally would, but make the 2 inner layers 100% coverage ground planes, instead of routing ground as traces. It’s like a big blanket of insurance for cross talk and signal integrity. 4 layer boards are so cheap these days there’s not a good reason to avoid them, in my opinion.
4 Likes
The statement about op amps is correct! Where bandwidth has improved with the Gen3 modules, it’s because of the switch to active attenuverter and switch circuits (the video signal never goes through a passive resistive element like a potentiometer or toggle switch.) So it would be correct to say that the modulation bandwidth (whenever the signal goes thru an attenuverter) has been improved, but the base signal path is almost identical.
Seeing folks get rid of useful modules in their rig because the maker doesn’t specifically say it’s “HD ready” without any qualification of what that really means has been sad to see.
Agreed, this is the exact opposite of what I wanted. I was trying to introduce HD timing options for sync and output – it’s about tempo, not resolution! – and I was trying to do it specifically in a way that existing systems could grow, not be replaced.
4 Likes
Prototype PCBs ordered from JLCPCB tonight! Taking bets on what stupid design mistake I’ve made this time…
3 Likes
So now my 4HP Soft Key prototype is being manufactured I’m turning my thoughts to the design of full Soft Keyer with an integrated VCA and wanted to gather some opinions on functionality.
I’d like to include an internal rectifier circuit to provide an outline function but was wondering about instead just including an ‘effects loop’ style input and output where any function could be patched into the key signal path, any preference?
I’m basing the VCA circuit on the Cadet Fader VI circuitry around the LT1251 (because of its snap to zero functionality), this is a 2Q VCA, I can’t see any benefit to having the option of switching between 4Q but wanted to see if anyone else had any suggestions.
1 Like
The effects loop i/o is a pretty novel idea. My vote is for that.
Were you planning on using the VCA to fade between a “foreground” and “background” image or were you thinking of using it as a fade-to-black multiplier?
I would just recommend 2Q.
1 Like
Great idea with the fx loop, best of both worlds - VCA is there but can patch in-between if you want it. BTW JLCPCB do aluminium PCBs now and they are a very good price. I just got back an order for black PCB with white silkscreen, perfect for front panels
3 Likes
Prototype PCB is here, will build and test this weekend, please keep your collective forum fingers crossed for me!
I’m aware that the front panel is pretty dull and the pots and jacks need better spacing on the next version, I just wanted to keep the main PCB to under 100mm to keep the cost down.
9 Likes
A tracing job after my own heart! Minimal vias show me that you put in a lot of care.
Did this end up being 4-layers with internal ground planes?
1 Like
May I see what your reference(s) buffer looks like in the schematic?
You may not want to populate those electro caps if they’re on the output of the TL072, but I want to have a look at it before making that statement.
Thanks so much, means a lot! Yes 4 layer with internal ground planes.
1 Like
Here’s the reference section, what do you think? I don’t actually use the 1V reference in this circuit, but put it in anyway because I had the spare Op Amp and will use it in the Keyer with integrated VCA.
1 Like
Ah-hah, I see. I suggest that you do not populate C15 and C16 but filtering the reference is important.
10uF may be too high a captivate load for the TL072 to drive. Instead, I would suggest placing small caps in parallel with R38 and R36. The cutoff may be calculated by:
fc = 1/(2piRC)
Where “R” is your feedback resistor. R38 and R36 on your board.
Check out this AC analysis:
And here is an example Lars has shared:
Another observation is the labelling of your 78L05’s pinout. It shows output and input switched, but maybe your silkscreen is correct.
Is the 7805 used for powering any IC’s? If not, a TL431 or LM4040 reference may be better choices. If the 7805 is used for power, then it should be fine.
2 Likes
Thanks Fox these are really helpful observations. You’re right about the regulator being backwards, thankfully should be fine to flip on the board and switch in the next revision.
Hadn’t considered the capacitive load issue, will add your suggestion in the next version. I really appreciate the insight, analogue electronics design can be such a minefield!
3 Likes
I would consider TL431 or LM4040 – if you are using the 7805 maybe a good idea to have a trimmer to set the reference. R35 & R36 I would make 10K resistors. Smaller resistors are fine but it’s more current loading that you need to have in this case. I also like to have a cap in series with feedback resistors (R38 and R36) to slow down any noise coming in.
2 Likes
Just general design notes on using voltage references: On a design like this, it’s not a big deal because the references are only used a couple of places. But in high gain scenarios where the VREF is distributed across multiple circuit blocks, you have to be very careful that you’re not getting any cross talk between the vref and the gain stages. So it can be better to have localized vref buffers. I ran into this problem on early prototypes of Staircase.
2 Likes
Testing this evening, just scoping things out at low frequency with non-video op amps, just incase I let the magic smoke out of any chips, so it’s a 50p mistake not a £7 mistake. Things are looking good at the moment!
8 Likes
Update - can thoroughly confirm version 1 does not work! Can’t yet confirm why it doesn’t work hahaha
Gain sections and clipping seem to be working but I’m getting no output. Frustratingly the way I’ve laid out the op amps is great for the pcb layout but makes isolating sections more difficult.
Next step is to chop the board up in a more aggressive manner to isolate each sub-circuit and hopefully chase down the issue.
10 Likes
Shame you didn’t nail it first time but I’d bet a “hole-in-one” is a rare occurrence even amongst very skilled designers once a circuit involves a certain amount of op-amps and a variety in pots & jacks.
Hopefully it’s not U6 causing the issue. This kind of narrow footprint caused me an issue with one of the Castle modules, thankfully it came to life after the 4th reflow & clean up.
Thanks for keeping us up to date with your progress @phosphenes &
you’ll figure out the issue quickly 
4 Likes