"MUSIC, formerly known as Music Group, a holding company encompassing Behringer, Midas, Klark Teknik and, most recently, the TC Group portfolio of pro audio brands, has announced its proposal to dismiss around 70 staff members at its manufacturing and office facility in Coatbridge, Scotland. In this event, the facility’s Tannoy loudspeaker production efforts will be transferred to MUSIC’s new manufacturing plant in Zhongshan, China; research and development as well as marketing activities will be relocated to MUSIC’s Manchester, UK offices. Approximately one year ago, Music Group acquired TC Electronic, TC-Helicon, TC Applied Technologies, Tannoy, Lab Gruppen and Lake, which comprised the TC Group. Tannoy has been based in Coatbridge for over 35 years. "
I would (and have proposed elsewhere ) that you use conventional OPS bias comp schemes for 1st order correction and then a uController for 2nd order correction.
I have currently settled on using conventional comp + NTC to provide a 2nd correction curve at higher temps.
Either of these two approaches offers very tight Ibias compensation.
Seems using a controller to provide 1st order bias comp is overkill - and in Schiit's case they don't seem to have cracked it.
I have currently settled on using conventional comp + NTC to provide a 2nd correction curve at higher temps.
Either of these two approaches offers very tight Ibias compensation.
Seems using a controller to provide 1st order bias comp is overkill - and in Schiit's case they don't seem to have cracked it.
In other news, a weasel tripped the LHC today! Large Hadron Collider: Weasel causes shutdown - BBC News
Won't do any good for the sales of the big hornloaded Coaxials to Japan if not made by guys in brown coats called bert!
Yeah and higher order compensation should have even longer time constants, where an algorithmic/lut system might be the easiest way forward rather than clever circuits. Then the microprocessor can run an lcd (because glowy lights are cool) in addition to some clever power-up and SOA/protection monitoring (fan control would be nice, too). Oh, hey! We've got some smart people working on something like that here at diyaudio.
But none of that would materially show up in sweeps on an AP...
Incompetent design. It's nice to dial back the bias to avoid thermal run-away but there is no reason to dial it all the way back to class C because that is what it is.
If the bias appears to run off, just dial it back to the design value.
Jan
until you try to have a peer-to-peer conversation - they really seem to just want to talk down to their marks/fanboys with "techy talk", at least by the evidence of my attempts at head-fi
I'll give you a good reason: a SoC with integrated A/D and D/A floating between the output stage gates, reading the Re voltage, calculate the bias current, implement a nonlinear square law control loop to the output stage bias voltage. Net result: a non-switching output stage that pretty much leaves the crossover distortions history (and also the dreaded gm doubling region for bipolars). The idea is to keep the non-switching condition on the fly, at HF or any transient conditions (not only at DC and LF as many simple "non-switching" output stage topologies do).
Not a big deal to implement with a $5 SoC and a little software, some decent CPU and signal acquisition/processing speeds are required (trivial today), what I find very difficult is to determine the stability of the nonlinear control loop and it's interaction with other local and global feedback loops.
So, 40-0-40DC, OMG !! Just a 3 pair sanken OPS will run 2 paralleled 4R subs
with that low voltage. You can literally short the output stage with that 16A DC SOA at 40V.
The new 5 pair Sanken will burn toaster wire <1R@50V.
You have no idea ..... Display's , I2C datalogging , and even mobile controlin addition to some clever power-up and SOA/protection monitoring (fan control would be nice, too). Oh, hey! We've got some smart people working on something like that here at diyaudio.
is being "hatched" now. I'm "old school" , and am just concerned with
reliable DC/thermal/current/inrush with scriptable parameters.
PS - I have fan control if I want it , just changing the script can port
a thermal "event" to the fan (instead of shutdown). At least on mine ... I never
overheated.
OS
Basically
. I know it is senseless to build something just because I have transformers to use up, but what the hell. N+1 is the correct number of amplifiers to own, where N is the current number
As long as the analog bias control is primary.
Then use the uPC to "fine tune" it. (What if the processor fails?).
To go full AB2 would need a fast uPC - https://www.arduino.cc/en/ArduinoCertified/IntelGalileo .
You said -
I've seen member vzaichenko's new non-switching EF3 - a (really)good THD reduction
at 20Khz as well as LF. Analog can achieve this quite easily.
I think the best hybrid is to keep analog as primary and use the uPC to
do "housekeeping" (thermal FB + remote adjustment/logging).
As you said, the effect of a digital control loop within the analog output
stage might be quite unpredictable.
OS
Last edited:
You've seen nothing when it comes to new $5 SoC's, Arduino is a pre-school kids slow toy
. Speed is the last concern.I've not seen that EF3 non-switching you are mentioning, but I certainly doubt you can implement true non-switching (with a square law, that is, keep the collector/drain output devices currents product constant for all frequencies and output levels) without a nonlinear feedback loop, be it analog or in the digital domain. Which makes both as difficult to deal with. I recall Edmond Stuart did an analog implementation, based on the LT1166 Linear Technology IC topology, unfortunately impaired by such stability issues.
The question is if such a digital bias schema is really required. I am pretty confident it won't have a iota of audible impact over any decent designed switching or non switching bias, while the sighted review chimps will only be appalled by this digital intrusion. So no, it could be just another technical challenge of the "because I can" type.
You've seen nothing when it comes to new $5 SoC's, Arduino is a pre-school kids slow toy
I recall Edmond Stuart did an analog implementation, based on the LT1166 Linear Technology IC topology, unfortunately impaired by such stability issues.
The question is if such a digital bias schema is really required. I am pretty confident it won't have a iota of audible impact over any decent designed switching or non switching bias, while the sighted review chimps will only be appalled by this digital intrusion. So no, it could be just another technical challenge of the "because I can" type.
That's (my link) a 400Mhz Intel atom arduino.
Edmund also did the AB2 with discrete's. Even combined EC ...
Auto Bias part II
I built them ..... still like the simplicity/stability of a fast EF3.
(below) is the NS-OPS.
http://www.diyaudio.com/forums/soli...e-old-ideas-1970s-ips-ops-14.html#post4648520
I agree that some of this is "because I can". I can't hear any error with a plain
ol' EF3.
NS (and my plain ol' output stage) both use them big Sanken's.
The "NS" does shave several PPM's off of the load testing , I'm sure this would
get the "review chimps" hard. It is a tradeoff between the PPM and a few hot
extra current sources (and more complexity).
OS
Attachments
Arduino IS a toy, always was. Just a very useful toy when you want something easy to program for a little home project.
- Status
- Not open for further replies.