bmoregnr, you have just about everything that is still known about the GD sound system. The JC-ULD module was specially made for the GD by Mark Levinson and operated at +/- 24V and could drive 600 ohms easily.
Thank you John, I learned a lot about it today so I now know everything it did. You saying JC-ULD module really helped open a lock however and I found these.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
That was on ebay in 2012 and the sales pitch was nice/interesting: “This amp is a product of the extreme engineering demands that were a standard of the bar set during the Wall of Sound generation and systems thereafter. I commissioned John Curl, renowned audio design engineer and one I consider the finest, to come up with a hybrid device that would do the job of line driver and be able to have front end flexibility. These are now in the all time great sounding device hall of fame. Audiophiles and engineers covet these for the outstanding sound quality. Even the DIY squad is trying to synthesize these incredible amplifiers. A casual search will reveal everything including app notes. Here is a chance to score a new, never been used version. I carried these as spare parts on the road but the number of them that were in service never failed so these were never used. All pins are in perfect condition. The JC on the label is John Curl; search him also.”
These Wall of Sound components I am researching always seem to turn out smaller and simpler than I imagined, so perhaps the summing amp was housed in one of the boxes as the bottom of that middle mic; which seems to make sense with it having a line driver.
An externally hosted image should be here but it was not working when we last tested it.
Thank you everyone for indulging me and I will now get out of your hair.
I have worked through more of these systems than I can remember now. The usual form is something like this:
Wireless module, at least Bluetooth and currently some WiFi flavor. I'm recommending Libre because they will work with smaller manufacturers. Streamium is in a middle ground but will require a significant upfront investment as well. PlayFi, AllPlay, etc. all require a major commitment and cash to play. They won't talk until the numbers exceed the entire audiophile market. They just don't make sense if you are building 1000 speakers. They all become practical at 100,000 speakers. This is really because the unit costs are small ($10-$40 ea.) so you just can't pay for engineering support for a handful of parts.
DSP SOC that handles digital, analog and I2S inputs. Usually with a sample rate converter on the input.The internal DSP can implement a number of biquads and limiter/compressor functions + crossovers and multiple I2S outputs or drive signals for a digital power stage.
Finally a digital power amp. The current generation is very good. While there is some cachet in a traditional amp for the output its not demonstrable in the sound until you get into the stratosphere, and then its questionable. DBT is a bad thing in these parts but really useful when real money is at stake.
Thank you for all the help demian. The ADAU 1701 seems to be very well suited for what I'm trying to do. I know it doesn't have FIR filters but I guess the IIR should be good for the xo and FR correction. I like that there is built in delay so time alignment is there. I'm still looking for a Class D amp solution that can make some real power at less than the 10% distortion rating they all seem to use in the data sheets. I'm not sure how you derate the power output for 1% or less distortion, compared to any class AB amp things look pretty grim on that count.
A bit of a memory jog, I hope she enjoys it there. (I did the Syntel boards😉)
https://accelconf.web.cern.ch/accelconf/e94/PDF/EPAC1994_2333.PDF
Op went badly, very badly!
Are these going to be offered for sale? I am interested in a complete system in a speaker for the Bedrooms and other rooms that are to small for a full system...
Marce,
Yes this is supposed to end up as a consumer product. If it was just for myself I would just buy a commercially available solution or these days use a PC with software to do all of the electronics. I'm working on a change to the grill I originally proposed and at the same time getting costs for all the speaker sections of the design like tooling costs and parts production costs. The electronics have been the biggest roadblock to finishing the design. Ignore the grill, here is the look otherwise of what I am working on. For a reference these speakers are 16" high x 6 1/2" wide and 14" deep. Finding that person who can put the electronics together on a nice surface mount board with all these disparate integrated circuits has been my quest now for awhile. So much talent here on this thread and this site and lots of help but that is still up in the air.
Yes this is supposed to end up as a consumer product. If it was just for myself I would just buy a commercially available solution or these days use a PC with software to do all of the electronics. I'm working on a change to the grill I originally proposed and at the same time getting costs for all the speaker sections of the design like tooling costs and parts production costs. The electronics have been the biggest roadblock to finishing the design. Ignore the grill, here is the look otherwise of what I am working on. For a reference these speakers are 16" high x 6 1/2" wide and 14" deep. Finding that person who can put the electronics together on a nice surface mount board with all these disparate integrated circuits has been my quest now for awhile. So much talent here on this thread and this site and lots of help but that is still up in the air.
Attachments
Last edited:
Yep. Rev B, page 16, column 2. Thanks Ed.
Urban myth says it does only IIR.
I have asked in the past if it has the potential to program FIR filters, I have got no answer.
Searching a bit deeper, there is some possibility
George
Scott and Simon thanks, I guess I bought the urban myth that there was no FIR capabilities with that chip. This makes me even more excited that I am getting close to a final solution. I'm just not convince about the usable power output of most of the class D amplifier outputs as they all list 10% distortion numbers at max ratings. I really want something that tops out at 100 watts into 8 ohms with low distortion like is so easy with a class ab amplifier. The numbers I'm seeing just look like hype so far. Why would I attach a high distortion amplifier to the design, makes no sense. There has to be a way to do that with low distortion using class D. A 100 watt amp for the bass/mid and 10 watts on the tweeter should produce a clean 107db output with a good 3db of headroom at that level. I know most people will never approach that loud but the speaker is easily capable of that sound level. The oval speaker is equivalent to a 7" driver in emissive area, sd, with much greater linear excursion than any other speaker I know of. I need the power and FIR filtering so I can do a Linkwitx transform to bring the bottom end up otherwise the enclosure is too small in internal volume for a standard box tuning. FR will be 35hz to 22Khz. In a larger box with no network attached I am measuring a FR bandwidth from the cone of 25hz out to 10khz. Crossover will be at 2.5Khz
Thank you for the approval George, I get all kinds of reactions. I hate the comments why don't I make it look like someone else design, I don't want to copy someone else design appearance. The lines between sections will be removed and it will look like one monolithic piece. Four of the screws around the dome tweeter are now gone, just the four in the middle have to remain. The connectors on the back are just a conceptualization of what is needed.
Thank you for the approval George, I get all kinds of reactions. I hate the comments why don't I make it look like someone else design, I don't want to copy someone else design appearance. The lines between sections will be removed and it will look like one monolithic piece. Four of the screws around the dome tweeter are now gone, just the four in the middle have to remain. The connectors on the back are just a conceptualization of what is needed.
Last edited:
From what I’ve gathered so far using FIR for filtering at low frequencies (say below 300hz) is very inefficient (number of instructions and data memory the filter needs is high). I think you’ll settle for IIR filters there
https://minidsp.com/applications/advanced-tools/linkwitz-transform
Are the (e.g.) TDA7498 diagrams so ugly? You’ll need 100W (10% THD) for very short peaks only. At 80W THD is 1%.
http://www.st.com/web/en/resource/technical/document/datasheet/CD00244535.pdf
http://www.irf.com/product-info/datasheets/data/ir4301.pdf
http://www.irf.com/product-info/datasheets/data/ir4302.pdf
Then it's all about cost. You can have some very good class D modules (ICE, Bruno's, Hypex ect) but they are not cheap.
George
Last edited:
FIR needs memory, multiplication horsepower - a chip designed for IIR may be able to do a dozen tap FIR - but that's not getting you far
FIR must be as long as the period of any frequency you want to substantially affect, much longer is needed for good control of filter characteristics, XO tricks like linear phase speaker summation requires FIR taps to store all of the samples needed for any time/phase compensation
FIR must be as long as the period of any frequency you want to substantially affect, much longer is needed for good control of filter characteristics, XO tricks like linear phase speaker summation requires FIR taps to store all of the samples needed for any time/phase compensation
George,
I've looked at the Hypex and a couple other commercial class D amps and besides the cost, which are high, the physical layout just won't fit in the space I have allotted for the rear heatsink. If it wasn't for the heat and power supply size I would just go class AB with something like Lazy Cat has done with his improved VSSA type of design. Small but needing some serious heat dissipation.
It may take some combination of active filtering IIR and some additional filtering to get the bass up with a slope on the bottom end. I think it is possible to do that. Of course it takes additional amplifier power to boost the low bass but that seems to be the biggest problem. The phase response of the speaker is rather nice and the impedance curve doesn't demand anything complex, the FR is very well behaved with no corrections so I am not going to have to do a lot of corrections, just pushing up the bass. I like the fact that there is a time delay in the ADAU chip that will make alignment of a simple two way system rather simple. Since the cone driver is very well behaved the overlap of the two devices shouldn't cause any headaches. I even thought of using a waveguide to move the tweeter back and physically time align but that brings in the requirement to shape the FR of the tweeter when you do that. Electrical alignment seems the better solution.
I've looked at the Hypex and a couple other commercial class D amps and besides the cost, which are high, the physical layout just won't fit in the space I have allotted for the rear heatsink. If it wasn't for the heat and power supply size I would just go class AB with something like Lazy Cat has done with his improved VSSA type of design. Small but needing some serious heat dissipation.
It may take some combination of active filtering IIR and some additional filtering to get the bass up with a slope on the bottom end. I think it is possible to do that. Of course it takes additional amplifier power to boost the low bass but that seems to be the biggest problem. The phase response of the speaker is rather nice and the impedance curve doesn't demand anything complex, the FR is very well behaved with no corrections so I am not going to have to do a lot of corrections, just pushing up the bass. I like the fact that there is a time delay in the ADAU chip that will make alignment of a simple two way system rather simple. Since the cone driver is very well behaved the overlap of the two devices shouldn't cause any headaches. I even thought of using a waveguide to move the tweeter back and physically time align but that brings in the requirement to shape the FR of the tweeter when you do that. Electrical alignment seems the better solution.
Thank you jcx for clarifying.
Attachment is taken from:
The Scientist & Engineer's Guide to Digital Signal Processing | Education | Analog Devices
Yes, apart from cost, size and heat generation are thinks you have to consider.
Don’t hesitate to test the class D chip solutions before deciding.
Get some of the ready made boards from e-bay and test them yourself under your application conditions (use a good adjustable, stabilized bench PSU).
Try both class D chip boards as well as class B chip boards.
This will save you a lot of time and guessing. Your choices will narrow down pretty soon.
With your loudspeaker, time alignment won’t be much of an issue but a wavequide will offer it’s other benefits too.
George
Attachments
Last edited:
Thank you for the advice again George. You always know where to look. There is just so much information on the AD site sometimes I miss something. A short waveguide would give me a chance to control the directivity but it almost always comes at the cost of a sloped response curve. I'm thinking a combination of active and some a passive network creating a slope in the bass section could emulate the Linkwitz function to extend the bass response. I just wonder how complex the phase could become, might make more problems than if would solve. The IIR filtering is probably going to be sufficient to work down low just to create a slope or step function. I'll read the AD guide. I'll look on Ebay, it's just that I see entire threads from people trying to make some of those boards actually work the way they are supposed to.
I'll see of I can get in contact with Kgrlee, he was one of the first to implement IIR in commercial designs, just he's been more than busy it seems on a project.
I'll see of I can get in contact with Kgrlee, he was one of the first to implement IIR in commercial designs, just he's been more than busy it seems on a project.
Last edited:
I have not been as impressed with the ST solutions as others. This TI part TAS5756M sounds very good, even driving Quad 63's. On the latest generation of amps they work very similar to traditional feedback amps with low distortion up to clipping. The 10% distortion numbers are a burden from the past but essential marketing terms. The customers are not sophisticated.
For a complete product there are many considerations and some will make a mess of others. Keeping the supply under 28V and using BTL amps plus optimizing the drivers to match the amps makes a big difference in the total product. I hate adding unnecessary DA AD conversions.
For a complete product there are many considerations and some will make a mess of others. Keeping the supply under 28V and using BTL amps plus optimizing the drivers to match the amps makes a big difference in the total product. I hate adding unnecessary DA AD conversions.
Demian,
There are were just so many choices on the TI website, most 2 channel, a few single channel, some analog and some PWM. Without having firsthand knowledge of any of them it got rather confusing to make a selection. I also looked at the ST chip, they didn't look very powerful from what I saw. There are plenty of companies to choose from but that doesn't make it any easier. The fact they all reference 10% distortion makes me think they were all originally aimed at car audio.
There are were just so many choices on the TI website, most 2 channel, a few single channel, some analog and some PWM. Without having firsthand knowledge of any of them it got rather confusing to make a selection. I also looked at the ST chip, they didn't look very powerful from what I saw. There are plenty of companies to choose from but that doesn't make it any easier. The fact they all reference 10% distortion makes me think they were all originally aimed at car audio.
- Status
- Not open for further replies.