Wynand said:
I agree with 'quasi', soft-start is normally done in the primary, mainly for large amplifiers (500W+) using toroidal mains transformers (conventional transformers are usually OK). I've never seen it done in the secondary?, anti-surge fuses are all that's required there! - although a primary soft-start would make the secondary softer as well.
May I suggest that you have a look at our LPS30 power supply?
It is a high quality complete power supply with inrush current limiter, etc. Althogh it was designed for our series of Class-D amplifiers it can work for other kinds of amps, of course.
http://www.coldamp.com/opencms/opencms/coldamp/en/productos/LPS30
Best regards,
Sergio
It is a high quality complete power supply with inrush current limiter, etc. Althogh it was designed for our series of Class-D amplifiers it can work for other kinds of amps, of course.
http://www.coldamp.com/opencms/opencms/coldamp/en/productos/LPS30
Best regards,
Sergio
After talking to some other people I've decided to scrap my schematic and opt for quasi's shematic.
It's way easier!! 😉
Thanks everybody.
It's way easier!! 😉
Thanks everybody.
AndrewT said:
Andrew,
I'm a little puzzled by this comment. In the circuit I suggested, the relay coil resistor is chosen so that the relay coil only sees about 18 V at full PSU rail voltage.
i.e. 63 V / (1100 + 450(coil resistance)) = 40.6 mA (A 24 V relay's nominal operating current is about 53 mA)
40.6 mA * 450 = 18.27 V
The 1.1k series resistor dissipates .0406^2 * 1100 = 1.8 Watts.
Neither the relay coil, nor the 10W W/W resistor run hot.
Regards,
Graeme
Hi Gra,
sorry I lost you.
The RC timeconstant on the relay coil takes time to charge up.
Roughly, the voltage on the cap will reach about 0.6times maximum in about one time constant. After about 5 time constants the cap will have charged to about 99% of maximum. However you have a relay hung on the end and this lengthens the time constant considerably. Now it might take 10 time constants to charge to 90% of maximum.
If the relay will pull in at 90% of maximum then the limiting resistors come out of circuit and start cooling.
BUT what if the mains input voltage is low? the relay may never pull in and the resistors will burn out because they are not designed to cope with mains loading.
Solution: lower the time constant by reducing the series resistor feeding the cap and relay to ensure the relay pulls in reasonably quickly when incoming voltage is at the low value specified by your power company.
Hang on, what if the mains is now running at nominal voltage? First the relay pulls in much more quickly, Good. But the cap keeps on charging to maximum over the next minute or so. Now the relay coil is taking more than nominal voltage all the time the amp is powered up. And if you are listening when incoming voltage is higher than nominal, just like what happens when you decide to settle down for a good listen before bedtime. Now your relay is seriously overheating.
There is a solution but few bother to include it. Use a transistor to switch the relay but add a cap across the relay transistor combination. The cap fires the relay with a high trigger current and the relay holds in on much reduced voltage, so keeping it cooler. Another advantage the RC time constant is unaffected by the relay because the transistor switch keeps it out of circuit until it triggers.
Edit:-
Is you're 63V the final PSU voltage? Then the problem gets much worse. While the limiting resistors are in circuit the PSU voltage is low, maybe 10 to 20% low (50V to 57V) leading to an even lower cap/relay voltage requiring a lower series dropping resistor to the relay. Now it all starts going seriously wonky when the PSU is fully charged with the same dropping resistor feeding the relay.
sorry I lost you.
The RC timeconstant on the relay coil takes time to charge up.
Roughly, the voltage on the cap will reach about 0.6times maximum in about one time constant. After about 5 time constants the cap will have charged to about 99% of maximum. However you have a relay hung on the end and this lengthens the time constant considerably. Now it might take 10 time constants to charge to 90% of maximum.
If the relay will pull in at 90% of maximum then the limiting resistors come out of circuit and start cooling.
BUT what if the mains input voltage is low? the relay may never pull in and the resistors will burn out because they are not designed to cope with mains loading.
Solution: lower the time constant by reducing the series resistor feeding the cap and relay to ensure the relay pulls in reasonably quickly when incoming voltage is at the low value specified by your power company.
Hang on, what if the mains is now running at nominal voltage? First the relay pulls in much more quickly, Good. But the cap keeps on charging to maximum over the next minute or so. Now the relay coil is taking more than nominal voltage all the time the amp is powered up. And if you are listening when incoming voltage is higher than nominal, just like what happens when you decide to settle down for a good listen before bedtime. Now your relay is seriously overheating.
does my question now make sense?
There is a solution but few bother to include it. Use a transistor to switch the relay but add a cap across the relay transistor combination. The cap fires the relay with a high trigger current and the relay holds in on much reduced voltage, so keeping it cooler. Another advantage the RC time constant is unaffected by the relay because the transistor switch keeps it out of circuit until it triggers.
Edit:-
Is you're 63V the final PSU voltage? Then the problem gets much worse. While the limiting resistors are in circuit the PSU voltage is low, maybe 10 to 20% low (50V to 57V) leading to an even lower cap/relay voltage requiring a lower series dropping resistor to the relay. Now it all starts going seriously wonky when the PSU is fully charged with the same dropping resistor feeding the relay.
Andrew,
The voltage at the relay coil after it has pulled in measures just over 19 V, at a supply rail voltage of 63 V. So the running current is just over 40 mA, which is only 80% of the coil's nominal rating. In my case, the current limit resistor in the transformer primary has almost no impact on the PSU voltage, as the amp's quiescent current is very low. It's only there to limit the maximum current at switch on caused by the toroidal transformer inrush and the initial charging current demand from the PSU smoothing caps.
In my own circuit, I used 22 ohms as the series resistor in the transformer primary, which limits the peak inrush current to about 15 A. In reality it will be lower due to transformer impedance etc. The inrush current lasts for a fraction of a second, so the resistor will take it quite happily. Even if the mains voltage drops so low that the relay doesn't close, the 22 ohm 10W resistor will take 674 mA at it's rated wattage. At 674 mA of primary current, the amplifier will have to be pulling close to 150W! Resistor overheat is nigh on impossible.
Incidentally, I tried the NTC thermistor approach at first. Even at ear shredding volumes the thermistors didn't even get warm, so I knew they weren't working and opted for the relay approach.
I'm satisfied that my relays are working at all times well within their operating limit, and would advise anyone who wants to use the relay option to make sure that they have the necessary understanding and a decent DVM to check that the relays are running within their safe operating limits.
The voltage at the relay coil after it has pulled in measures just over 19 V, at a supply rail voltage of 63 V. So the running current is just over 40 mA, which is only 80% of the coil's nominal rating. In my case, the current limit resistor in the transformer primary has almost no impact on the PSU voltage, as the amp's quiescent current is very low. It's only there to limit the maximum current at switch on caused by the toroidal transformer inrush and the initial charging current demand from the PSU smoothing caps.
In my own circuit, I used 22 ohms as the series resistor in the transformer primary, which limits the peak inrush current to about 15 A. In reality it will be lower due to transformer impedance etc. The inrush current lasts for a fraction of a second, so the resistor will take it quite happily. Even if the mains voltage drops so low that the relay doesn't close, the 22 ohm 10W resistor will take 674 mA at it's rated wattage. At 674 mA of primary current, the amplifier will have to be pulling close to 150W! Resistor overheat is nigh on impossible.
Incidentally, I tried the NTC thermistor approach at first. Even at ear shredding volumes the thermistors didn't even get warm, so I knew they weren't working and opted for the relay approach.
I'm satisfied that my relays are working at all times well within their operating limit, and would advise anyone who wants to use the relay option to make sure that they have the necessary understanding and a decent DVM to check that the relays are running within their safe operating limits.
Hi Gfinlayson,
seems like your set up is thoroughly checked and works at all voltages.
Can you post some details with relay model number and resistor/cap values?
Did you manage to measure the minimum pull in voltage of the relay? and also the max drop out voltage?
seems like your set up is thoroughly checked and works at all voltages.
Can you post some details with relay model number and resistor/cap values?
Did you manage to measure the minimum pull in voltage of the relay? and also the max drop out voltage?
Andrew,
The relays are from Maplin.
The specs are 24 V DC -20% +10%.
They're industry standard units, used by the thousand on industrial control systems. In my experience, failure rates are very, very low.
Specs for service life are > 10 milllion operations mechanically, and >100,000 operations electrically at full load (10 Amps)
Minimum pull-in for mine is about 14.7 V, drop-out voltage is around 7.5 V.
When the amps have been running for a while, the coil voltage seems to settle out at about 20 V, well above minimum pull-in and well below nominal rating.
I used a 1000uF 35 V electrolytic with a 1k 10W W/W resistor on the inrush suppression and a 2200uF 35 V electrolytic with a 1k 10W W/W resistor on the speaker outputs. This keeps the speakers out of circuit until everything has settled, so no switch-on thump. 'Golden-eared' purists will argue that relays make an audible difference, but I'll be damned if I can hear it. These relays are way better than most of the ones you'll find in any commercially manufactured amp, which is frightening given their cost. The inrush relay is wired to the + 63 V rail, and the speaker relay is wired to the - 63 V rail to keep the transformer balanced as toroids have a reputation for humming when the secondaries are unevenly loaded.
Hope the info is useful.
Best Regards,
Graeme
The relays are from Maplin.
The specs are 24 V DC -20% +10%.
They're industry standard units, used by the thousand on industrial control systems. In my experience, failure rates are very, very low.
Specs for service life are > 10 milllion operations mechanically, and >100,000 operations electrically at full load (10 Amps)
Minimum pull-in for mine is about 14.7 V, drop-out voltage is around 7.5 V.
When the amps have been running for a while, the coil voltage seems to settle out at about 20 V, well above minimum pull-in and well below nominal rating.
I used a 1000uF 35 V electrolytic with a 1k 10W W/W resistor on the inrush suppression and a 2200uF 35 V electrolytic with a 1k 10W W/W resistor on the speaker outputs. This keeps the speakers out of circuit until everything has settled, so no switch-on thump. 'Golden-eared' purists will argue that relays make an audible difference, but I'll be damned if I can hear it. These relays are way better than most of the ones you'll find in any commercially manufactured amp, which is frightening given their cost. The inrush relay is wired to the + 63 V rail, and the speaker relay is wired to the - 63 V rail to keep the transformer balanced as toroids have a reputation for humming when the secondaries are unevenly loaded.
Hope the info is useful.
Best Regards,
Graeme
- Status
- Not open for further replies.