Glide On Fade Awesome Mixer finally finished pics too!

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Congrats Dave! Your patience alone is worthy of much praise. If I was wearing a hat right now, I'd take it off to you. :thumb:
 
fum-

thanks for the link! Im gonna try it on a session and see how big of a deal the crosstalk actually is in use. Changing the buss impedance with a resistor iin series to the makeup amp seems like an easy enough way to tweak it out though.

[quote author="analag"]Go Dave!!!!!
So when can I come over and listen to it[/quote]

Id love to meet up if you'd like to come down, Im in NJ. Im free up until thanksgiving then have a session which should take me through until mid january, so before or after whatever works. Would love to hear some of the amps you've designed for sure.

dave
 
No, the resistor doesn't go in series with the input of the amp, it has to go in parallel with the input, between the "+" and "-" input terminals.

For the bussing, is this more or less the arrangement you ended up using--as discussed in an earlier thread?
GIF

As for the meter buffer and the related subject of how those old "decibel" meters differ from a VU meter, I'll start a separate thread about it on the Drawing Board, rather than cluttering up this thread.
 
Cool, I just wanted to make sure I was on the right page. In that case, try a 180-ohm resistor across pins 4 and 5 of the 325s. This will drop the buss level by 20dB and should improve the crosstalk by an equivalent amount. The gain of the 525 will also have to be increased by 20dB to maintain the same output level, but that shouldn't be a problem because 49dB is available. S/N will definitely worsen but whether or not it'll be audible remains to be heard.

If you don't have 180: a 150, 200 or 220 are close enough, especially for an experiment. Don't be surprised if the mix buss isn't completely to blame for the crosstalk--I think the daisy-chained arrangement of the card connections could have something to do with it as well.

As for the meter buffer and the related subject of how those old "decibel" meters differ from a VU meter, I'll start a separate thread about it on the Drawing Board, rather than cluttering up this thread.

Ah, f*ck it. A few pictures can tell the story just as well:

Photo Album
 
[quote author="NewYorkDave"]If you don't have 1.8K: a 1.5K, 2K or 2.2K are close enough, especially for an experiment. Don't be surprised if the mix buss isn't completely to blame for the crosstalk--I think the daisy-chained arrangement of the card connections could have something to do with it as well. [/quote]

I'll try that in the coming weeks and see where it puts me. Changing the value of that resistor simply changes the level drop? If thats the case, might be cool to dial in the right compromise between cross talk and makeup noise with a pot.

I think bussing the power like I did was probably the biggest mistake I made on this one.

dave
 
Congratulations on finally getting this project done, Dave. It looks great. Seems like you were bugging me for those input boards forever, and when I finally got them to you, the box was finished in no time.
Before you fuck with the circuit, listen to the crosstalk. What does it sound like? Is it mostly treble, mostly bass, or is it broadband? If it's treble, it's capacitively coupled. That seems unlikely due to the space between your channels. If it's bassy, it could be inductively coupled (transformer leakage, also unlikely) or it could be coupled through the power supply (very likely, and fairly easy to fix). If it's broadband, it's due to the passive summing, and this is the most likely scenario. You can either add a shunt resistor as discussed, or you can increase each channel's series resistance, or both. NYD's suggestion sounds about right for solving the problem with a single resistor per buss, and should be easy enough to implement.
You could also rewire the 325 card into an inverting arrangement so that you have a virtual ground input which should pretty much eliminate any crosstalk. But then your "passive" summing is now "active" summing. Pretty minor distinction, isn't it?
 
Wouldn't ya know it? I was one decimal place off in my mental arithmetic in my last post. :oops: Of course, this didn't dawn on me until I'd powered off the computer for the night and had gone to bed. Anyway, I corrected the post: it should have read 180 ohms where it said 1.8K.

You can, as you surmised, try a variable shunt--and, when you find the most acceptable compromise between crosstalk, loss and noise, replace it with a fixed resistor. Here's one to try:

variableshunt.jpg


I would not recommend increasing the buss loss beyond 40dB because at that point, you're probably hovering perilously close to your noise floor.

To JUM: I was going to suggest going to active summing, too, but if I recall correctly that would entail a complete redesign of the master fader arrangement Dave is using. Dave, please correct me if I'm wrong.

I believe the bussed power and grounds could be a source of at least some of the x-talk. Taking the V+, V- and power and signal grounds from each card to one central point would probably improve matters. Adding R-C decoupling to the power feed of each card is highly recommended; you could even tack the components on the card connectors if you wanna give it a whirl without ripping the mixer apart. As a matter of fact, you might wanna try that before you even start playing around with the mix buss. It "couldn't hoit" since decoupling is just good practice anyway. Use 33 ohms in each leg (which will drop only 2V or so at max current draw) and at least 1000uF/25V.

api_decoupler.jpg
 
all good point to consider. damn, just when I thought I was done, haha.

Im gonna try it out on my next job and see if Im complaining about something on the scope or on the speakers. I think rewiring the power bussing is probably the smartest thing I can do. The very first thing I did on this project was wire up the XLR's and buss the connectors thinking way back when that I could never screw that up. Well, if you look at the input/output numbering, I screwed that up and the power bussing Id bet is not helping me across 8 channels... Its amazing what you can learn in two years. Or not learn, haha.

cj, trade you one mixer for a pile of plexi transformers.

dave
 
I am sending you a plexi API output.
400 turns instead of 286, which means twice the henries, on a 75/25 core, with Ni on the corners for max fringe reduction.
Hopefully, you won't like it!
 
im almost afraid to ask but, what is the waight on that thing? i wouldnt expect you planing on giging with with it.

Great job btw.
 
didnt use the step bit, they are hard to use for exact holes I think. I have two diffferent ones and they both leave a rounded edge to the hole which isnt really good for mounting IMO, especially with XLRs. Others seem to like them though, maybe I just bought cheaper ones.

It actually doesnt weigh as much as you'd think, its not much heavier than the neve box I just finished which has 4 1272 channels it. Either way, I didnt build it for the road, just to sit at the studio.

dave
 
all the electros on each channel, mostly power decoupling. Wound up with BC, panasonic, Nichicon and a mix of both on the buss amps.

dave
 
You know, to be honest, Im still not over the shock of it actually working, Ive been burning it in, still waiting for the puff of smoke... I cant wait to start the next one with neve amps, winter project...

dave
 
Back
Top