Yes, it's another soft-start design! 🙄
This one is meant to be as fail-safe as possible, both for the amp and the soft-start circuit itself; it's relays, and inrush resistor. I'm employing a lot of Boolean, pinball-machine transistor/relay/optocoupler logic here; something I am not very fluent in, so it's been a week of obsessing and staying up till 6AM trying to figure it out! :headbash All the "ANDs" and "NOT's" and "ORs"...and the "If such-and-such a condition is TRUE, then don't NOT do the THING"...
ugh it hurts my brain.
But I think it works!
Download "The Spice" and tell me if I am missing anything.
Hoppe's Brain GFA-555 Soft-Start
The various voltage sources are set up to simulate normal operation. To make things abnormal, tweak their parameters and you can test the logical functions of the circuit. For example, you can test for chattering AC condition by pulsing the 24V unregulated supply voltage. You can simulate the completion of capacitor charging by pulsing off the voltage seen at the optocoupler that monitors inrush current.
Commercial disclaimer: This is intended for power supply boards for the Adcom GFA-555 that I will be producing and selling on my website. However I am happy to share the schematic for anyone to use or modify. I hope it's a good design that people find useful.
One issue I wanted to address, is the possible destruction of the inrush resistor or relays, if there should be a fault condition such as a brown-out or chattering AC input. Many designs do not reset the relay timer quickly enough when power is lost, and when power returns, the relay stays on and there is no soft-start action. This one resets in a couple milliseconds.
I wanted to get away from the assumption that just because time has passed since the power was applied, that the capacitors are charged and it's safe to click in the bypass relay. This condition is merely inferred! Why not test to see if this condition is true before clicking in the relay? I found a few circuits on the web that monitor power supply capacitor voltage, and then close the relay when they are mostly charged. However, this still doesn't get to the root of the issue, which is...
The relay should not click in until the current through the inrush resistor is low. Time is not the issue, nor is power supply capacitor voltage. It's the current through that resistor, indicated by the voltage across that resistor. When that is low, it's safe to pull in the bypass relay.
This design monitors a handful of conditions that guide it's operation. (Opto-couplers are neat-o!)
Both the power relay and the soft-start relay have one-second time constants, but this is nothing to do with waiting for the capacitors to charge. The timers only serve to limit relay chattering to 1Hz. Both timer's trigger states are held logic high, by default when power is lost, and will not start the timer unless conditions are met.
Here are the fault conditions I am trying to handle, do you see anything else I should be considering?
What am I missing here? There must be something!
Do I need to pepper some 10nF compensation caps around the switching transistors, or will they be OK? I haven't used a high-side MOSFET switch before, I like it! Is 470K alright for a pullup? I am trying to make the remote power sense optocoupler very sensitive so it will operate decisively, and I don't want the MOSFET to ever be unsaturated.
This one is meant to be as fail-safe as possible, both for the amp and the soft-start circuit itself; it's relays, and inrush resistor. I'm employing a lot of Boolean, pinball-machine transistor/relay/optocoupler logic here; something I am not very fluent in, so it's been a week of obsessing and staying up till 6AM trying to figure it out! :headbash All the "ANDs" and "NOT's" and "ORs"...and the "If such-and-such a condition is TRUE, then don't NOT do the THING"...
ugh it hurts my brain.But I think it works!
Download "The Spice" and tell me if I am missing anything.
Hoppe's Brain GFA-555 Soft-Start
The various voltage sources are set up to simulate normal operation. To make things abnormal, tweak their parameters and you can test the logical functions of the circuit. For example, you can test for chattering AC condition by pulsing the 24V unregulated supply voltage. You can simulate the completion of capacitor charging by pulsing off the voltage seen at the optocoupler that monitors inrush current.
Commercial disclaimer: This is intended for power supply boards for the Adcom GFA-555 that I will be producing and selling on my website. However I am happy to share the schematic for anyone to use or modify. I hope it's a good design that people find useful.
One issue I wanted to address, is the possible destruction of the inrush resistor or relays, if there should be a fault condition such as a brown-out or chattering AC input. Many designs do not reset the relay timer quickly enough when power is lost, and when power returns, the relay stays on and there is no soft-start action. This one resets in a couple milliseconds.
I wanted to get away from the assumption that just because time has passed since the power was applied, that the capacitors are charged and it's safe to click in the bypass relay. This condition is merely inferred! Why not test to see if this condition is true before clicking in the relay? I found a few circuits on the web that monitor power supply capacitor voltage, and then close the relay when they are mostly charged. However, this still doesn't get to the root of the issue, which is...
The relay should not click in until the current through the inrush resistor is low. Time is not the issue, nor is power supply capacitor voltage. It's the current through that resistor, indicated by the voltage across that resistor. When that is low, it's safe to pull in the bypass relay.
This design monitors a handful of conditions that guide it's operation. (Opto-couplers are neat-o!)
- The 12VDC power supply is good. (And it's safe to operate the relays with sufficient force.)
- AC power is good (Inferred by DC power supply to the soft-start being good)
- Either the power switch is on, or the remote trigger is on
- The inrush resistor is not overheated
- The current through the inrush resistor is low, and therefore the capacitors have charged.
- 1 second has passed with all parameters OK before either relay will click in. (Chattering is limited to 1Hz)
Both the power relay and the soft-start relay have one-second time constants, but this is nothing to do with waiting for the capacitors to charge. The timers only serve to limit relay chattering to 1Hz. Both timer's trigger states are held logic high, by default when power is lost, and will not start the timer unless conditions are met.
Here are the fault conditions I am trying to handle, do you see anything else I should be considering?
- Brown out, voltage low: DC power supply will be less than 10.6V and so power trigger is held high and timer will not start
- Brown out, voltage on threshold of keeping DC power supply OK: Power relay may start, but it will momentarily drag supply voltage down, causing relay to click out again. This repeats at 1Hz. I need to physically build the circuit to see if the relay actually pulls in, or if it resets before the contacts actually touch. It could be a problem if it makes brief momentary contact over and over again.
- Brown out, power fluctuating: Relays will both cut out when power goes below a certain level. The relays will click in if power comes up again, and they may cycle in and out, but at a maximum rate of 1Hz.
- AC power chattering: Power relay timer self-resets and will not start until power is steady. If the power remains on for longer than 2 seconds, both the power and soft-start relay will click in, and both will reset if power is lost again.
- Remote signal or power switch chattering: Power timer resets and will not start.
- Capacitors NOT charging up, or something is wrong with the amp and causing excess current draw: Power relay will come on, soft-start relay will not. Inrush resistor will heat up, and eventually 130C thermostat will open, causing everything to turn off. (Power resistor can handle 270F, so the thermostat should open well before that, even if it is rapidly heating.)
What am I missing here? There must be something!
Do I need to pepper some 10nF compensation caps around the switching transistors, or will they be OK? I haven't used a high-side MOSFET switch before, I like it! Is 470K alright for a pullup? I am trying to make the remote power sense optocoupler very sensitive so it will operate decisively, and I don't want the MOSFET to ever be unsaturated.
Last edited:
Mmmkay, well, to answer a few of my own questions...
Yes, some compensation caps will keep the base current spikes reasonable when this thing clamps the timing caps to the rail, or when it cuts out from low supply voltage, or when the first stage fires up the second stage. These caps slow everything down a bit, but the timing capacitor reset is still a maximum of 1.6ms.
Also, I didn't need to be driving the bases so hard just to drag the 555 threshold line high and low, so I increased the value of the base resistors. Conversely, I increased the base current to the low-voltage detector transistor, so that it will act more decisively around the zener voltage.
And I think I can predict what will happen in a brown-out situation, where the DC supply voltage dips out when the power relay tries to click in. The simulation says there will be a 2ms blip trying to turn on the relay before the supply voltage dips enough for the reset to kick in. I think that will be no problem, and the relay will not make momentary contact. According to the relay datasheet, it's guaranteed to switch on in a maximum of 15ms, so I don't think there's any way it's going to move much in just 2ms. I will test this physically once I bread-board the thing. The coil's inductance will appear open-circuit, momentarily, at switch-on, so that might stretch out this response. I will have to check this on the breadboard.
The new Spice file is here...Hoppe's Brain GFA-555 Power Board II
Yes, some compensation caps will keep the base current spikes reasonable when this thing clamps the timing caps to the rail, or when it cuts out from low supply voltage, or when the first stage fires up the second stage. These caps slow everything down a bit, but the timing capacitor reset is still a maximum of 1.6ms.
Also, I didn't need to be driving the bases so hard just to drag the 555 threshold line high and low, so I increased the value of the base resistors. Conversely, I increased the base current to the low-voltage detector transistor, so that it will act more decisively around the zener voltage.
And I think I can predict what will happen in a brown-out situation, where the DC supply voltage dips out when the power relay tries to click in. The simulation says there will be a 2ms blip trying to turn on the relay before the supply voltage dips enough for the reset to kick in. I think that will be no problem, and the relay will not make momentary contact. According to the relay datasheet, it's guaranteed to switch on in a maximum of 15ms, so I don't think there's any way it's going to move much in just 2ms. I will test this physically once I bread-board the thing. The coil's inductance will appear open-circuit, momentarily, at switch-on, so that might stretch out this response. I will have to check this on the breadboard.
The new Spice file is here...Hoppe's Brain GFA-555 Power Board II
No comments? I hope that means I am doing it right!
Well I'm about to find out. I've ordered new boards.
I was going to proto-board the thing to test it, but I have already spent way too much time obsessing over this. LT Spice is probably not wrong about what it does. I'll just order boards and experiment with them. If there's something wrong I'll order a new batch. What a luxury to have such inexpensive PCBs!
There's been a few changes since my last post. I was doing the high-side mosfet switch a little wrong. I was pulling the gate to ground, but that is around 24 volts in this design, way too much VGS! So I changed the pull-up arrangement to a voltage divider so the VGS is always 12Vor less.
Here's the new SPICE.
Meanwhile, I have in-house, a simpler version of the soft-start board that I ordered before I started to get all fancy with the circuit. This one has a simple and dumb 555 timer like my old design. Works great though!
Testing with a 555 transformer and caps.
I like the way this thermostat worked out! It's mounted to the underside of the power resistor, and held there by the stand-off. It's a Sensata Airpax 67L130 in TO-220.
Well I'm about to find out. I've ordered new boards.
I was going to proto-board the thing to test it, but I have already spent way too much time obsessing over this. LT Spice is probably not wrong about what it does. I'll just order boards and experiment with them. If there's something wrong I'll order a new batch. What a luxury to have such inexpensive PCBs!
There's been a few changes since my last post. I was doing the high-side mosfet switch a little wrong. I was pulling the gate to ground, but that is around 24 volts in this design, way too much VGS! So I changed the pull-up arrangement to a voltage divider so the VGS is always 12Vor less.
Here's the new SPICE.
Meanwhile, I have in-house, a simpler version of the soft-start board that I ordered before I started to get all fancy with the circuit. This one has a simple and dumb 555 timer like my old design. Works great though!
Testing with a 555 transformer and caps.
I like the way this thermostat worked out! It's mounted to the underside of the power resistor, and held there by the stand-off. It's a Sensata Airpax 67L130 in TO-220.
Last edited:
So of course there were problems!
For one, I was getting relay chattering when the incoming AC voltage was hanging around the threshold detect voltage. At a very specific point near 95V, the relay would click in, but the voltage drop caused by that would cause the circuit to reset and immediately try to close the relay again. So I added about 0.7s hysteresis to the AC detect. (Through R20 and C21)
Also, I was using the secondary voltage of the transformer for the AC detect, and this has poor regulation, which was exacerbating the effect. I changed it to monitor the incoming AC directly.
Now, the chattering is limited to at least 0.7 seconds, and the relay contacts won't be damaged, and the amp won't mind either.
The only other issue I had, is that the cooling for the surface-mount LM7812 regulator was insufficient, and it was running around 60C, and it would be worse inside a warm amp. So I added a pretty large heatsink with a standard TO-220 LM7812.
It works great! Feel free to steal the circuit design, but please don't copy my actual board design.
Here's the Spice.
For one, I was getting relay chattering when the incoming AC voltage was hanging around the threshold detect voltage. At a very specific point near 95V, the relay would click in, but the voltage drop caused by that would cause the circuit to reset and immediately try to close the relay again. So I added about 0.7s hysteresis to the AC detect. (Through R20 and C21)
Also, I was using the secondary voltage of the transformer for the AC detect, and this has poor regulation, which was exacerbating the effect. I changed it to monitor the incoming AC directly.
Now, the chattering is limited to at least 0.7 seconds, and the relay contacts won't be damaged, and the amp won't mind either.
The only other issue I had, is that the cooling for the surface-mount LM7812 regulator was insufficient, and it was running around 60C, and it would be worse inside a warm amp. So I added a pretty large heatsink with a standard TO-220 LM7812.
It works great! Feel free to steal the circuit design, but please don't copy my actual board design.
Here's the Spice.