Okay, I found this, which seems to be the place to put that kind of stuff, but it looks mostly theoretical with few test results and I don't see a reference to ChocoHolic's version as discussed here. Any chance it could be added?
http://www.diyaudio.com/forums/class-d/104822-favorite-threads-posts-class-d-forum.html
http://www.diyaudio.com/forums/class-d/104822-favorite-threads-posts-class-d-forum.html
Workhorse had chosen such arrangement because of speed.
The collectors of Q6 and Q7 are more acting like a current source, rather than voltage source, so it is not problem to tie them to a fixed voltage.
Remaining downside of this comparator is that Q5 and Q9 can be driven into saturation which then will add pretty some delays.
...so if the specs are good and the design of this and other parts aren't too tricky, I might end up using "a few" of these.
Be ensured that building this amp is tricky
Also be ensured that the audiophile specs are complete overkill for an outdoor concert set. For your application you should choose a simple circuit which simply delivers the needed power and is trimmed to low average losses (mixed weighting of stand by losses and various load losses).
This crazy design which marries high end audiophile spec + high power is not a natural match for your application, but of course possible.
If you go for it, choose V1.3
Make sure that you really follow the 'get it going' instructions and have the amp running perfectly fine at current limited low power rails of +/-12V, before changing to the full power rails.
Only the operation with the current limited +/-12V rails will allow you to debug the building errors, without much danger of destruction.
Oohps - caught a single star rating during the last few days.
When tracking the votings I receive, for strange reasons there seem to be almost only 5 star or 1 star ratings. A single star rating indicates somebody is angry with me.
Definitely I would prefer to hear/read what's wrong rather than
just getting slapped.
When tracking the votings I receive, for strange reasons there seem to be almost only 5 star or 1 star ratings. A single star rating indicates somebody is angry with me.
Definitely I would prefer to hear/read what's wrong rather than
just getting slapped.
Yeah, I expected that. S/N isn't all that great in live sound anyway, let alone outdoors, which leaves lots of room to hide some distortion. But I do want it to sound at least somewhat like the original source.
My priorities are:
- Lightweight: No more than 50lb per full-range powered speaker stack + 50lb per NiMH battery pack. (all portable to multiple gigs per day in different locations)
- Efficient: Pretty much dictates Class D. (and horns if they don't violate #1)
- Loud: Feel the bass!!! Outside! More practically, use the batteries hard despite the efficiency.
- Low-ish THD+N: At least enough to not notice in a live concert.
I'm going the DIY route because I don't know of any commercial amps or powered speakers in that scale that aren't either massive (linear), expensive (switching), or annoying (auto-shutoff during background filler music). So, DIY it is. Plus #2 above = DIY Class D, and diyAudio is the only place I've found that even makes an attempt at tackling it. Perhaps the difficulties can actually be overcome now instead of simply brushed off "because muggles can't do it" like I read in so many other places.
Sure it's tricky, but if a company can build something, then so can a skilled amateur. 
It's like youtube, either they like it or they don't, because of number of reasons, that has nothing to do with content. In this context with you or the amp. Or amazon rating and such. You can get one star rating just because it is too complex to understand and "they" can't understand it/ build it... It's just a tool for feedback that is not 100% for product or project itself. (too complex, don't like that)
..muggles...
..that's a cool brush off.Where can I read such?
I wouldn't dare to sign this statement.
Nevertheless classD amps can be build by skilled amateurs.
Basically everybody has to decide on his own, whether he is confident to build the amp of this thread. For sure it is not simple and needs good building
skills and for debugging good electronic know how / expirience + scope & multimeter.
Looking to your list of requirements it would be more lazy to buy one of the cheap high power Behringer amps and substitute the mains power supply with a battery power supply.
..muggles...
Where can I read such?
That was my summary of several sites, including my favorite for analog audio: Elliott Sound Products - The Audio Pages (Main Index)
None that I remember actually used that term, unfortunately.
Notice I said "skilled". That means actually skilled and not just ignorant of the pitfalls/dangers. Ideally, companies are just collections of these people that have agreed to share a common goal, and they only happen to be paid for it.
What I was getting at is that the laws of physics don't change just because you're part of a company or doing it yourself.True, but it would toast the warranty on something I don't have a clue about. If something's going to break in the field, I'd much rather have a shot at hacking it back into service until we get home.
In the company I am working for, these people only happen to combine their expertise and have access to equipment which amateurs simply do not have.
I can only talk for me - and have to admit, I have to expect the same for me even when my own design fails during the party.
Anyhow - if you have carefully read the builders pack and are confident that you can build and debug it... Welcome to the show!
In case of questions, of course I will give you remote support.
However due to living on different continents we cannot simply sit together at the bench for taming the beast.
Okay, so you're starting to get into one difference: they can pool their collective profits to buy fewer copies of each expensive tool than there are people, thus making nicer stuff affordable. Even so, there's a lot that can be done with used (monetarily depreciated but fully functional) equipment. And some equipment can be DIY'ed itself for much cheaper than retail without the unnecessary bells and whistles for the specific application. And if you're lucky, some companies (not all, but some) allow their equipment to be used for personal projects after hours as long as it remains useful to the company.
I can only talk for me - and have to admit, I have to expect the same for me even when my own design fails during the party.
Exactly. I finally found your files a few pages back in this thread, I've been through the schematics and ran your gerbers through GerbV just to have a look, and it doesn't actually look all that bad, except that your drill file is mirrored compared to the way I do it. It's conceptually simple, just sensitive to layout and a few construction details, which you've done already. I might just have to build one and see how it goes. I was contemplating whether to design my own from the power stage back and work through the issues as they come, but since you've already solved them, I think I'll learn from you instead. Either way, I'll actually know something about the final product when it's done.
Just to catch a potential bug before I buy all the wrong parts, it appears that the gerbers for v1.3 and v1.5 are identical according to my diff viewer, and a manual comparison of the schematics seems to be identical as well. Is that right? I didn't compare the BOM yet.
The gerbers are identical, schematic not.
PCB & Gerber were intended for V1.3.
For V1.5 I needed not just different component values, but also some additional components which are not foreseen in the PCB and have to be placed P2P.
-An RC at the output for carrier shaping in order to press k3 below -90db
-A voltage clamp at the output of U201 and some resistors/caps in the complex gain setting network.
All the components of V1.5 which were not foreseen in the PCB have the 'Add' in their name (Schematic and BOM).
Back to V1.3.
It has a lower switching frequency which allowed to set the overcurrent protection limitation above 50Apeak (while for V1.5 I have chosen a level short above 40Apeak).
In your application also the idle current consumption is relevant.
==> Choose a larger dead time by leaving open R314, distortion still remain better than adequate.
Some results with this are given in posting #396.
http://www.diyaudio.com/forums/class-d/224052-systemd_2kw-any-interest-open-design-40.html
You should also avoid unnecessary high current draw in the low power sections, so my proposal for the comparator would be MAX 913 and OPamp AD8620. But that's more fine tuning - feel free to look which types you can get - how much supply current they need - and make your choice.
From my experience with NiMH I can get roughly 60Wh/kg.
This would lead to 120cells 1.2V/10Ah in series, centertapped (+/-72V nominal, roughly +/-85 when fully charged).
Do you already have the cells on hand?
If not, but in case you intend to buy new cells, I would propose to have a look to LiFePO4.
This type of Li-battery is definitely not as fire critical as Li-polymere or most other Li-types used in notebooks or smart phones.
Energy/weight is somewhere between Li-polymere and NiMH.
90Wh/kg can be regarded as pretty realistic with LiFePO4.
Hm... seeing battery capacities of 10Ah and larger, it appears to be Ok, to get the auxiliary +/-12V rails from tapping between the cells (2x9cells of 1.2V). When using AD8620 + MAX913 you can expect approx. 40..50mA draw from the +12V and approx 30...40mA from the -12V
Assuming approximately 5h operating time the resulting discharge unbalance for the cells is less than common capacitance tolerances.
Only the Drv auxiliary should be provided with a seperate pack (13 cells of 1.2V), assume roughly 130-160mA current consumption here.
Alright! Good to see that the straight-forward documentation goes with the version that I actually want.
I haven't bought anything yet. I'll probably do this on a current-limited supply first and then build the batteries. So far, it looks like the best capacity/dollar is about 400 AA's from eBay...but that's going to be annoying to build a pack with.
My original idea was Lithium Polymer in radio control hobby style packs, for their energy density like you said. But then I looked at the regulations about flying with them. There's a limit to the amount of lithium that each person can carry on (forbidden in checked luggage) that equates to roughly 200Wh per person. For a typical team of 10-ish, that's 2kWh give or take a few people. And then we'd have to assemble them correctly. They don't seem to care at all about nickel batteries; just make good and sure that they can't discharge catastrophically.
I may end up with some auxiliary batteries to run the FET drivers and line-level signal processing (analog crossovers and corrective EQ's), but my original idea was a couple of DC-DC buck converters just to simplify the battery configuration. If all outputs are bridged, can I get away with a 2-wire 160V safety-interlocked battery connector and a capacitor-stabilized resistor-divider for the center tap?
I haven't bought anything yet. I'll probably do this on a current-limited supply first and then build the batteries. So far, it looks like the best capacity/dollar is about 400 AA's from eBay...but that's going to be annoying to build a pack with.
My original idea was Lithium Polymer in radio control hobby style packs, for their energy density like you said. But then I looked at the regulations about flying with them. There's a limit to the amount of lithium that each person can carry on (forbidden in checked luggage) that equates to roughly 200Wh per person. For a typical team of 10-ish, that's 2kWh give or take a few people. And then we'd have to assemble them correctly. They don't seem to care at all about nickel batteries; just make good and sure that they can't discharge catastrophically.
I may end up with some auxiliary batteries to run the FET drivers and line-level signal processing (analog crossovers and corrective EQ's), but my original idea was a couple of DC-DC buck converters just to simplify the battery configuration. If all outputs are bridged, can I get away with a 2-wire 160V safety-interlocked battery connector and a capacitor-stabilized resistor-divider for the center tap?
Theoretically it should be OK to use DC/DC buck converters for the auxiliaries, however I did not test such a set up and potential HF-noise interactions with this amp.
Capacitor staibilized resistor network for the center tap:
... also never tried, but from theory....
The capacitors are already in place (all the rail caps from +82V to GND and from -82V to GND).
The resistor divider to center the DC level of GND in the middle of both rails basically should work -however I have to look more detailed before I can propose values.
Capacitor staibilized resistor network for the center tap:
... also never tried, but from theory....
The capacitors are already in place (all the rail caps from +82V to GND and from -82V to GND).
The resistor divider to center the DC level of GND in the middle of both rails basically should work -however I have to look more detailed before I can propose values.
Got it. Thanks!
Thinking some more about batteries, and I'm not so sure anymore about having 170VDC available during transit, even if it does have a safety-interlocked connector. Maybe the same number of cells can be arranged into two series packs to drive the entire system instead of one isolated pack per side. Still 85VDC per pack, which you can't be flippant about either, but only half of what it could have been and I naturally get a center tap too. Add some bucked rails for processing and FET drivers, so I don't have to care which brick is on the positive or negative side, and I think I'm good for the FOH.
Also, what experience, if any, do you guys have with modern AC-designed gear on high-voltage DC? Specifically looking at a Behringer XR18 or similar. All of the open-source designs that I've seen for switching power supplies rectify the AC input immediately to HVDC and then buck that through a much smaller isolation transformer than would be required for 50/60Hz. I don't see any reason not to run that design on 170VDC, seeing that it typically does anyway with 120VACrms, but a commercial thingy with proprietary everything may or may not like it. Certainly don't try it with linear supplies; the input transformer will just burn up and not pass the DC.
Maybe I can just make a "modified sinewave" aka "cheater" inverter without the boost stage, seeing as I already have the required voltage and everything should be isolated anyway. (effectively two hot wires and no neutral) Then I can run pretty much anything from it.
Running an AC SMPS at DC typically bears three pitfalls:
1. The thermal design of the rectifier can be critical, because at DC
only two diodes are used and have to carry the current continously, instead
of the alternate sharing over all 4 diodes in AC operation.
2. EMI.
3. Fuse design. Most fuses are designed for AC, while DC is much more difficult to interupt.
All three topics depend on the very specific design, there is no general statement possible which would allow an obvious yes or no from outside view.
Last but not least - you never know which additional fancy feature has been implemented and might need the AC...
Personally I would not recommend it, except the manufacturer states also DC operation.
1. The thermal design of the rectifier can be critical, because at DC
only two diodes are used and have to carry the current continously, instead
of the alternate sharing over all 4 diodes in AC operation.
2. EMI.
3. Fuse design. Most fuses are designed for AC, while DC is much more difficult to interupt.
All three topics depend on the very specific design, there is no general statement possible which would allow an obvious yes or no from outside view.
Last but not least - you never know which additional fancy feature has been implemented and might need the AC...
Personally I would not recommend it, except the manufacturer states also DC operation.
usually PFCs are designed so that the current shape follows the voltage waveform up to a certain bandwith limit. Most PFCs can deal with triangle, square voltage etc. input, although the power factor may be lower and the threshold for input under- and overvoltage may be shifted due to confusion in the input voltage measurement circuit which is often just peak measurement.