There's the RNC, which is similar in that it has an analog signal path and a digital side chain. Not everyone's crazy about the RNC, but that may well be caused by its algorithms.
I wouldn't use an audio codec or any other S/D converter, as that will introduce enough latency to be audible on fast attack settings. A SAR ADC like the one integrated in most microcontrollers will do fine, or else a cheap external one like the $1 MCP3201.
JDB.
[working on a different shiny hammer -- a switchable hybrid between the SSL, GSSL and the THAT 2252 appnotes. Everything's digitally switchable through either the front panel or MIDI and you can even have two sidechains active at the same time. Now all I have to do is transform the breadboarded/P2Ped prototype to a PCB layout...but Real Work is cutting into my playtime lately]
new shiny hammer- or Universal comp
Re: new shiny hammer- or Universal comp
I did a product definition maybe 10 years ago while still at peavey for something like a x8 digital controlled analog comp.
That could have been price effective, sharing one set of pretty basic controls for 8 comps, but project never finished.
This new one is a little more optimistic and not really a value play. I still would like even more controls (or computer interface)
to allow me to fully model most classic comps.
The speed of the A/D in the micro families I work with are more than fast enough for side chain. In fact sensible LPF
in I/O path could allow side chain to be a little faster than main path, but I'm not inclined to go lower than 50kHz -3dB
for primary path..
JR
That could have been price effective, sharing one set of pretty basic controls for 8 comps, but project never finished.
This new one is a little more optimistic and not really a value play. I still would like even more controls (or computer interface)
to allow me to fully model most classic comps.
The speed of the A/D in the micro families I work with are more than fast enough for side chain. In fact sensible LPF
in I/O path could allow side chain to be a little faster than main path, but I'm not inclined to go lower than 50kHz -3dB
for primary path..
JR
Cancel the "cancel culture", do not support mob hatred.
- mediatechnology
- Posts: 5444
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: new shiny hammer- or Universal comp
jdb: Cool! We've discussed (elsewhere) using an rms feed-forward and FB SSL-style (>2 peak) compressor simultaneously using both VCA's Ec ports. One port for rms, the second for peak. Have you looked at using simul Ec?working on a different shiny hammer -- a switchable hybrid between the SSL, GSSL and the THAT 2252 appnotes. Everything's digitally switchable through either the front panel or MIDI and you can even have two sidechains active at the same time.
Roger is laying out a VCA "brick" that brings out both Ecs. I've almost carved up a Pico Comp proto a couple of times to try that. It's that work/playtime thing again. Ours wasn't as fancy though. MIDI recallability is very cool.
Re: new shiny hammer- or Universal comp
[Not trying to hijack anyone's thread here -- just say the word an I'll start a new thread]
I am reasonably happy about having two sidechains. Although I have implemented averaging, in my limited testing I prefer having the compressor pick the highest SC output (higher == more compression).
JDB.
[that higher == more compression bit me while breadboarding, as I'd forgotten that the sidechains were set to output negative voltages for connection to Ec-, and for a while I couldn't figure out why my analog max()-function didn't appear to do anything]
Yeah, and I'm not too eager to do that. As you know Ec influences the distortion trimming; I'm using the THAT-suggested way to feed some Ec into the trim port to lower distortion over a larger gain range. I know no easy way of achieving that when you're driving both control ports, and making the op-amp that presents a low impedance to Ec- accept differential signals is easy enough.mediatechnology wrote:Have you looked at using simul Ec?
I am reasonably happy about having two sidechains. Although I have implemented averaging, in my limited testing I prefer having the compressor pick the highest SC output (higher == more compression).
JDB.
[that higher == more compression bit me while breadboarding, as I'd forgotten that the sidechains were set to output negative voltages for connection to Ec-, and for a while I couldn't figure out why my analog max()-function didn't appear to do anything]
- mediatechnology
- Posts: 5444
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: new shiny hammer- or Universal comp
Do both! We'd love to have you start one, or two or...[Not trying to hijack anyone's thread here -- just say the word an I'll start a new thread]
A very good point.As you know Ec influences the distortion trimming; I'm using the THAT-suggested way to feed some Ec into the trim port to lower distortion over a larger gain range. I know no easy way of achieving that when you're driving both control ports, and making the op-amp that presents a low impedance to Ec- accept differential signals is easy enough.
FWIW The >2 response seems to be taking hold with the GSSL mod that Keith's done. Where I would typically ship 3X 2180s to a builder I'm now shipping 4.
Re: new shiny hammer- or Universal comp
Not to hijack my own thread but i finished my little console meter project and I really like working with the 16 bit PIC micro. Even though I could make a passable meter with the old 8 bitter (10b a/d) the 16b (12b a/d) was much nicer.
For jollies, I coded up a true RMS algorithm for the VU portion of my simultaneous peak/VU meter. It wasn't that hard.. I multiply the signal by itself to square it before I integrate it over time to simulate meter attack/release time constant's, then I compute the square root, before mapping the integration result to the LED display. The square root was a little tricky since the cheap micro has no square root function, and common algorithms for sqrt involve division. There is a divide instruction in the micro but it takes 19 clock ticks, per divide. I wrote my own sqrt algorithm to iteratively perform square root using the multiply instruction which is one clock tick per 16bx16b multiply... So I did the whole sqrt in less clock time than maybe two divides.
Not that it matters much.. in a side by side close inspection of one meter displaying true RMS with the other displaying simple average, I couldn't perceive any difference. nada...
Well at least now I know ho to do it if I want to be able to say RMS for my uber comp puppy side chain.
JR
For jollies, I coded up a true RMS algorithm for the VU portion of my simultaneous peak/VU meter. It wasn't that hard.. I multiply the signal by itself to square it before I integrate it over time to simulate meter attack/release time constant's, then I compute the square root, before mapping the integration result to the LED display. The square root was a little tricky since the cheap micro has no square root function, and common algorithms for sqrt involve division. There is a divide instruction in the micro but it takes 19 clock ticks, per divide. I wrote my own sqrt algorithm to iteratively perform square root using the multiply instruction which is one clock tick per 16bx16b multiply... So I did the whole sqrt in less clock time than maybe two divides.
Not that it matters much.. in a side by side close inspection of one meter displaying true RMS with the other displaying simple average, I couldn't perceive any difference. nada...
Well at least now I know ho to do it if I want to be able to say RMS for my uber comp puppy side chain.
JR
Cancel the "cancel culture", do not support mob hatred.
- mediatechnology
- Posts: 5444
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: new shiny hammer- or Universal comp
Very cool: So when can a buy a kit with a pre-coded PIC and PCB?
Curious: Is your squaring operation is double-precision or mantissa/exponent? How do you avoid dynamic range limitations?
Curious: Is your squaring operation is double-precision or mantissa/exponent? How do you avoid dynamic range limitations?
Re: new shiny hammer- or Universal comp
I'm starting out with 12 bit a/d and rectifying that which leaves me with 11 bit raw data.. Ieft shift it up to top of 16 bit word and square it I get a 32 bit result but I just use top 16bits... the final result is parsed to only 3dB steps displaying around a 50 dB range. Which is maybe 9 bits...It oooks like I have more dynamic range but have vetted that completely.
I can do a kick ass meter... the current one is pretty sweet.. 4x 16 led segments, actually using 6 a/d inputs, 4 audio channels and 2 solo/// the micro does all the solo meter switching with one simple logic line... It also has a few other tricks... and all mute for emergency situations, I even built in an auto calibrate...
I still want to use this engine for a digitally controlled analog VCA dynamics processor... I need to interface it with an Iphone or something clever so I can tweak all the parameters I want to /
If I do a meter kit, I'm tempted to make several versions,,, maybe one for the processor and N sundry variants for how many LED outputs... Or I guess the customer could decide how many LEDS they want to hook up. Play might be to sell a tiny postage stamp pcb with processor and little more, let the DIYer roll their own LED meter boards...
Kind of kind of like modern day digital LM3915 but sooo much hipper...
JR
I can do a kick ass meter... the current one is pretty sweet.. 4x 16 led segments, actually using 6 a/d inputs, 4 audio channels and 2 solo/// the micro does all the solo meter switching with one simple logic line... It also has a few other tricks... and all mute for emergency situations, I even built in an auto calibrate...
I still want to use this engine for a digitally controlled analog VCA dynamics processor... I need to interface it with an Iphone or something clever so I can tweak all the parameters I want to /
If I do a meter kit, I'm tempted to make several versions,,, maybe one for the processor and N sundry variants for how many LED outputs... Or I guess the customer could decide how many LEDS they want to hook up. Play might be to sell a tiny postage stamp pcb with processor and little more, let the DIYer roll their own LED meter boards...
Kind of kind of like modern day digital LM3915 but sooo much hipper...
JR
Cancel the "cancel culture", do not support mob hatred.
- mediatechnology
- Posts: 5444
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: new shiny hammer- or Universal comp
The new shiny hammer connection.I still want to use this engine for a digitally controlled analog VCA dynamics processor... I need to interface it with an Iphone or something clever so I can tweak all the parameters I want to /
I like the meter concept John and the flexibility. One thing Roger and I learned from the original 3914-based GR meter we did was that too many build options created support. That meter could be built either GR or level, constant V or I LED regulation etc. When shipped with a Pico Compressor the meter would generate all the questions. Yours would seem to be much simpler in the pick-n-place department with software controlling the modes. If you did do a kit or something you might consider an LED assembly vs. discretes. We're always having to correct backwards LED insertion.
Re: new shiny hammer- or Universal comp
I have real work to consider but my desire would be to probably make a finished small pcb with maybe a handful of parts around the micro. The basic module would be set up for 2 audio inputs and 2 serial digital outputs. Pushing the data out serial would allow more flexibility after the fact.
For example right now I am writing 64 bits of LED on/off data in one long bit string and then latching the whole deal into 8 serial in parallel out SR. In my digital 3915 I would use two different latch lines one for L and the other for R. I could push something like 16 or 24 bits of Left data, then latch left side, push 24 bits of right data then latch right. These shift registers can be daisy chained in series. By pushing the lower level LEDs first and latching after the top 8 leds, one could customize their display by how many latches and LEDs they hang on the output. If they connect only one latch the get top 8 leds, two latches top 16 leds, 3 latches all 24 leds. The extra output resolution is there to use or not. One could even imagine modular LED PCB with two 8 bit latches and 16 leds to cobble together a stereo array.
Personality pins on the micro could select between peak/ave/both, bar/dot, linear/3 dB log, AC/DC.... So like a 3914/15/16 on steroids in one chip. One can extend the meter in 8 LED increments but you don't have to populate all leds so you can build any length display up to 24. I can even bring out the a/d resistor string so you can custom tweak the range (within the modest 3.3V rail. The micro can interface with 5V logic for latches so more than adequate to drive LEDs.
So not unlike the 3915 pop this mini module in the middle of semi custom meters. The PCB would cost more than micro... Maybe if the part is available in through hole, I could let DIY make the whole PCB... While layout will matter for 12 bit A/D in 3.3V rail.
later
JR
For example right now I am writing 64 bits of LED on/off data in one long bit string and then latching the whole deal into 8 serial in parallel out SR. In my digital 3915 I would use two different latch lines one for L and the other for R. I could push something like 16 or 24 bits of Left data, then latch left side, push 24 bits of right data then latch right. These shift registers can be daisy chained in series. By pushing the lower level LEDs first and latching after the top 8 leds, one could customize their display by how many latches and LEDs they hang on the output. If they connect only one latch the get top 8 leds, two latches top 16 leds, 3 latches all 24 leds. The extra output resolution is there to use or not. One could even imagine modular LED PCB with two 8 bit latches and 16 leds to cobble together a stereo array.
Personality pins on the micro could select between peak/ave/both, bar/dot, linear/3 dB log, AC/DC.... So like a 3914/15/16 on steroids in one chip. One can extend the meter in 8 LED increments but you don't have to populate all leds so you can build any length display up to 24. I can even bring out the a/d resistor string so you can custom tweak the range (within the modest 3.3V rail. The micro can interface with 5V logic for latches so more than adequate to drive LEDs.
So not unlike the 3915 pop this mini module in the middle of semi custom meters. The PCB would cost more than micro... Maybe if the part is available in through hole, I could let DIY make the whole PCB... While layout will matter for 12 bit A/D in 3.3V rail.
later
JR
Cancel the "cancel culture", do not support mob hatred.