Thanks for the prompt reply… (seem to have sparked something off here). Actually sensing the supply rails directly was going to be my next query.
I'm no engineer but my guess work suggests that this idea would not only stop DC from getting to the output (saving the speaker), but also save the amp… or most of it anyway. Also it would enable us to get rid of the supply rail fuses… I think… no?
Plus, could we do without the output relay as well? While arranging for the DC offset detection to trigger rail shut-down, cutting off the DC at source, even if it's derived from DC being fed into the input?
Whenever I come across something I don't understand I always like to try and understand it (even if it's only for academic reasons). So in view of this, I'm still curious about how emitter resistor current sensing would be wired. Is it like this? (hopefully to pic has shown up).
Attachments
Your diagram is correct.
Theoretically you could remove the output relay for protection reasons, but it's still needed for turn on and turn off thump delay. Many amps put out some ugly offset on power up and power down too.
It would likely be good to leave the rail fuses in place as added protection. The current sensors will react much faster than a fuse, so they would either need to be set to trigger at a higher current or slowed down to prevent false triggering. It would be better to leave them as fast sensing devices and let the fuses take care of slower failure events. The ultimate scheme would be to connect them to analog inputs and write software to allow high current for short bursts, and lower for longer lengths of time, but this would add a lot of work to the processor and would likely slow it down.
Theoretically you could remove the output relay for protection reasons, but it's still needed for turn on and turn off thump delay. Many amps put out some ugly offset on power up and power down too.
It would likely be good to leave the rail fuses in place as added protection. The current sensors will react much faster than a fuse, so they would either need to be set to trigger at a higher current or slowed down to prevent false triggering. It would be better to leave them as fast sensing devices and let the fuses take care of slower failure events. The ultimate scheme would be to connect them to analog inputs and write software to allow high current for short bursts, and lower for longer lengths of time, but this would add a lot of work to the processor and would likely slow it down.
Hi,
jwilhelm:
What you want to do it is what I do in my amplifier. Since I controlled/regulated the rails voltages with the micro I do not used output relays. What I do it is ramp up/down the rails voltage and at the same time monitoring the speakers output current if the current it is too high I shutdown the rails voltages. Same when it is running. If the current in one of the channel is high I shutdown the rails voltages and last if needed shutdown the AC. You can use the ACS78xx before or after the speaker but you will need two one for the pos rail and one for the neg rail. Installing it in the output you will need only one since the ACS78xx read neg/pos current.
jwilhelm:
What you want to do it is what I do in my amplifier. Since I controlled/regulated the rails voltages with the micro I do not used output relays. What I do it is ramp up/down the rails voltage and at the same time monitoring the speakers output current if the current it is too high I shutdown the rails voltages. Same when it is running. If the current in one of the channel is high I shutdown the rails voltages and last if needed shutdown the AC. You can use the ACS78xx before or after the speaker but you will need two one for the pos rail and one for the neg rail. Installing it in the output you will need only one since the ACS78xx read neg/pos current.
This system is meant to be connected to any amplifier, so there's no way of knowing how the amplifier will start or shut down. I prefer to keep the output relay for that reason. I also prefer the idea of monitoring the current at the amplifier input. It's possible to have an amp oscillating and suffering shoot through without having high output current. I would prefer to catch this and shut it down before it lets the stink out.
Actually you can use rail disconnect fets as current resistors, with optoisolators as sensors. Trip point 1.2v/.04ohm=30 amps trip point. I've bought some .041 max resistance nfets. This current should be adequate for most 3 to 5 output transistor pair amps. You can connect the phototransistor in open collector OR function with other fault detectors to set the fault latch, and turn off the drive current to the rail fets driver. I've got the photovoltaic rail fet drive IC input in series with the green side of the fault LED.
If you need more sensitive rail current fault, you can put 0.1 ohm or 0.2 ohm resistor in series with the rail nfets and put the photodiode of the isolator (plus current limit resistor) across both. Or buy a higher resistance nfet.
If you need more sensitive rail current fault, you can put 0.1 ohm or 0.2 ohm resistor in series with the rail nfets and put the photodiode of the isolator (plus current limit resistor) across both. Or buy a higher resistance nfet.
Last edited:
I've got the rail fet between mains cap and the output transistors. So I don't see the mains caps having anything to do with the response time. My PV-1.3k has .1 uf out locally on the output transistor board rail, that is all the stored charge out there.
If Rds is .0025, then use a .04 ohms resistor in series with it to set off the optoisolator. Better really,a hard resistor value doesn't decrease with temperature the way a a nfet resistance will. Optoisolators are about 10 mhz devices, so are 74hc74, so that should be fast enough for anybody. A Panasonic APV1122 would take longer to discharge the gate charge of the rail fet to shut it off, unless you use optoisolators in series with the red fault led to short out fet gate to source (which I'm doing)
If Rds is .0025, then use a .04 ohms resistor in series with it to set off the optoisolator. Better really,a hard resistor value doesn't decrease with temperature the way a a nfet resistance will. Optoisolators are about 10 mhz devices, so are 74hc74, so that should be fast enough for anybody. A Panasonic APV1122 would take longer to discharge the gate charge of the rail fet to shut it off, unless you use optoisolators in series with the red fault led to short out fet gate to source (which I'm doing)
I'd prefer just to stick with the hall effect current sensors. There are versions with a programmable alarm set point that would make signaling problems to the control board simple, and at the same time giving a current reading for future data-logging options. This and some rail shut downs could easily be connected in series between the supplies and amplifiers.
I'm bouncing around the idea of doing a couple through hole layouts of the latest versions of the protection system. I'll likely never use these myself because the boards would be larger but if others would like to use them it would be worthwhile. Can people let me know here if they would be interested in this?
My project is still in design and I'm looking at protection, so yes I'd be very interested the 'latest version'.
Can I ask if the through-hole aspect is the only difference from the previous version? Is this a new version with a possible hole-through option? Will the new version include supply rail sense/shut-down for example? Also, while I'm here, is there anything wrong with using SSRs for AC mains switching?
Through hole versions would be a little different than the SMT for programming, otherwise they are pretty much the same. I can't find Serial to USB UARTs in through hole, so I would likely make provisions to plug an inexpensive CP2102 adapter from Ebay or Amazon for programming, with the option to remove the microcontroller and plug it into a Uno. All the same add ons will still be able to connect to the boards.
SSRs would work for the power relays. I've been avoiding them due to space requirement and cost. On boards power relays are cheap and compact. These systems get expensive with the amount of parts involved, so I've been trying to stay with inexpensive options where possible, but not compromising where it really counts, such as solid state speaker relays.
SSRs would work for the power relays. I've been avoiding them due to space requirement and cost. On boards power relays are cheap and compact. These systems get expensive with the amount of parts involved, so I've been trying to stay with inexpensive options where possible, but not compromising where it really counts, such as solid state speaker relays.
Would you be needing a system for a single main supply transformer or duals?
I will publish an updated v1.2 document in the first post soon, describing the current finished, "production" version
Hi Valeriy,
Any chance of us getting those documents of the latest version in first post?
It is long promised I am currently building your smt version, I have found building with smt components isn't as bad as I thought. The only thing I didn't like is soldering the atmega, the pitch is way too small to solder each pin individually. The only way is to bridge all the pins and then use solder wick to draw off the excess solder.
I also almost messed up by uploading the i2c scanner sketch with my isp programmer. Thankfully I stumbled upon why that is a bad idea before it was too late. Uploading a sketch with an ISP programmer overwrites the boot loader. That would make programming impossible when the board was fully populated due to other components interfering with the ISP interface. If I understand correctly, the atmega can only be programmed through the serial interface once the board is fully populated, and you need a boot loader to do that.
So if you don't plan on making an updated through hole version, at least consider a version with a Dip atmega. That would be helpful to a micro controller novice like myself.
I wouldn't worry about a built in USB interface for a through hole version, TTL serial interfaces are fairly easy to come by.
If you scrub the board with isopropyl alcohol first, then apply liquid rosin you can solder the Atmega328 quite easily with a little solder and a big hot iron. Clean is the key. Solder adheres better, so it won't blob as much. A little practice with these and they become quite easy. I've been soldering with a cheap hot air rework station and some solder paste lately. I though it would blow parts around, but it works great and the surface tension of the solder aligns the parts perfectly, just like in an oven.
The Arduino IDE won't let you load a sketch through the USBTinyISP. You would need to do this in AVRDude. If you do manage mess up the bootloader, it can be reflashed easily if you remove the voltage loss detector first.
The through hole version will likely be set up to accept a CP2102 off Amazon or Ebay for easy programming, but the Atmega328 ran easily be removed and put in a Uno board for programming too.
Here's a YouTube video of an easy way to solder SMT ICs. https://www.youtube.com/watch?v=5uiroWBkdFY
The Arduino IDE won't let you load a sketch through the USBTinyISP. You would need to do this in AVRDude. If you do manage mess up the bootloader, it can be reflashed easily if you remove the voltage loss detector first.
The through hole version will likely be set up to accept a CP2102 off Amazon or Ebay for easy programming, but the Atmega328 ran easily be removed and put in a Uno board for programming too.
Here's a YouTube video of an easy way to solder SMT ICs. https://www.youtube.com/watch?v=5uiroWBkdFY
Last edited:
I never got back to the oven yet. The display wouldn't light up on the controller, but I've never had the time to look into it yet.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.