Ignore the previous message then.
You opened softstart.asc which is not the ntc version - see above.
The traces are at the top of the trace viewer. So the red is from U1, the relay. More specifically it is the node that connects when the relay trips. So blue trace shows the current through R3 i.e. the soft start in operation, and the red is when the relay trips and bypasses the resistor bank. Vrelay rises like that as a result of the RC. In the hysteresis models you'll see it jump.
You opened softstart.asc which is not the ntc version - see above.
The traces are at the top of the trace viewer. So the red is from U1, the relay. More specifically it is the node that connects when the relay trips. So blue trace shows the current through R3 i.e. the soft start in operation, and the red is when the relay trips and bypasses the resistor bank. Vrelay rises like that as a result of the RC. In the hysteresis models you'll see it jump.
BTW ntc-20130313.lib is a library from EPCOS for their NTC. I don't think I ended up using it in the end. Rather I used a generic NTC model and populated the relevant NTC parameters (RO, B, Rth, TAU) from the Ametherm data sheet.
Andrew, may I ask you about NTC's on the transformer secondary side...the circuit advice from SG says to use four NCT's ...two before the rectifier bridge (the AC side) and two after the rectifier bridge ( the DC side).
If one uses resistors as current limiters on the transformer primary side...say 25R/50W... is it then necessary to use NCT's on the secondary side?
If yes....(I would like to keep everything as cool as possible inside the amplifier cabinet)....would it be a good idea to bypass the NCT with a relay after the delay period?
regards Gudmund
I consider the two situations as being quite different.
Starting a transformer
Charging a capacitance.
Starting a transformer:
When starting a transformer, there is a large current flowing around the PRIMARY circuit and NONE around the secondary circuit.
This can use a "soft start" to limit the primary current. This has a two fold advantage.
a.) allows one to start the transformer using a close rated fuse.
b.) allows other equipment around the house to not be subjected to a voltage glitch during start up while the high current pulses surge into the primary.
The current limiter can be a fixed resistor, or bank of series connected resistors, or a power NTC, or a bank of series connected Power NTCs.
I don't like paralleled power resistors for soft start duty. The wire, or trace, inside the high value resistor is thin and not very robust. The starting duty is severe!
One must NEVER parallel connect Power NTCs used for soft start duty.
Charging the capacitance:
The capacitors begin to charge AFTER the transformer has started.
The capacitors may be current capability limited. Starting from uncharged with a high charging voltage leads to very high initial charging current. This can exceed the maximum current allowed for the capacitor and/or can exceed the dI/dT rate of change allowed for the capacitor.
Slow charging can be implemented to limit both dI/dT and Imaxcharge.
Again fixed resistor or Power NTCs can be used. But in this case power NTCs are particularly suitable. look at the datasheet and apnotes of the Power NTCs. It appears to me that they were designed for this particular duty. To do this job properly the NTC is placed in series with the capacitor to be charged. That's the way the manufacturers show it in the datasheets. They do not show a transformer between the NTC and the capacitor. It does not matter what side of the NTC that you place the rectifier. They are all inseries. The current MUST flow from the secondary and flow through the capacitor and the rectifier and the NTC and then return to the secondary. As always there is a circuit and the current must RETURN to the Source. Use twisted pairs.
In both slow charge and soft starting there is a considerable advantage to using a bypass across the limiting resistance after the purpose has expired. Low source impedance.
For soft starting the time period can be as few as 5 cycles of the mains frequency, but most recommend around 100ms to 300ms.
For slow charging the time period is determined by the RC time constant, but can be around 5s to 30s. The variable NTC resistance makes estimating the RC time constant not much better than guesswork. Since the NTC is effectively self protecting it can be left unbypassed for relatively long periods without risk of damage. I would aim for the longer end of the 5s to 30s period.
Relays can be used. A 556 can be used to trigger the two different time delays.
Starting a transformer
Charging a capacitance.
Starting a transformer:
When starting a transformer, there is a large current flowing around the PRIMARY circuit and NONE around the secondary circuit.
This can use a "soft start" to limit the primary current. This has a two fold advantage.
a.) allows one to start the transformer using a close rated fuse.
b.) allows other equipment around the house to not be subjected to a voltage glitch during start up while the high current pulses surge into the primary.
The current limiter can be a fixed resistor, or bank of series connected resistors, or a power NTC, or a bank of series connected Power NTCs.
I don't like paralleled power resistors for soft start duty. The wire, or trace, inside the high value resistor is thin and not very robust. The starting duty is severe!
One must NEVER parallel connect Power NTCs used for soft start duty.
Charging the capacitance:
The capacitors begin to charge AFTER the transformer has started.
The capacitors may be current capability limited. Starting from uncharged with a high charging voltage leads to very high initial charging current. This can exceed the maximum current allowed for the capacitor and/or can exceed the dI/dT rate of change allowed for the capacitor.
Slow charging can be implemented to limit both dI/dT and Imaxcharge.
Again fixed resistor or Power NTCs can be used. But in this case power NTCs are particularly suitable. look at the datasheet and apnotes of the Power NTCs. It appears to me that they were designed for this particular duty. To do this job properly the NTC is placed in series with the capacitor to be charged. That's the way the manufacturers show it in the datasheets. They do not show a transformer between the NTC and the capacitor. It does not matter what side of the NTC that you place the rectifier. They are all inseries. The current MUST flow from the secondary and flow through the capacitor and the rectifier and the NTC and then return to the secondary. As always there is a circuit and the current must RETURN to the Source. Use twisted pairs.
In both slow charge and soft starting there is a considerable advantage to using a bypass across the limiting resistance after the purpose has expired. Low source impedance.
For soft starting the time period can be as few as 5 cycles of the mains frequency, but most recommend around 100ms to 300ms.
For slow charging the time period is determined by the RC time constant, but can be around 5s to 30s. The variable NTC resistance makes estimating the RC time constant not much better than guesswork. Since the NTC is effectively self protecting it can be left unbypassed for relatively long periods without risk of damage. I would aim for the longer end of the 5s to 30s period.
Relays can be used. A 556 can be used to trigger the two different time delays.
Last edited:
Thank you for your detailed explanation.
The delay by guess for the inrush current resistor was 1 second, and for the inrush NCT about 5-6 seconds. I can extend the delay time for the NCT.
What would you recommend as inrush resistor value by 2 x 500Vac transformers?
I have thought about 25 -30 R ......according to your advice, we shall test with two NCT's (in line the plus and minus rails) between the rectifier bridge and the capacitor supply.
Regards Gudmund
For soft start I would go very much shorter than 1s, I use around 200ms.
That keeps it permanently cold even if repeated restarting is required.
ONE 500VA 230Vac transformer should be close rated fused @ around 2A, but you may find that T1.6A works for years. Or you may have to go up to T2.5A if your voltage is high and the primary inductance of the transformer is low.
The nominal maximum current will be around 500/230 ~ 2.2Aac . There's that T2A fuse rating.
If I double this number to 4.4A, that has given me a good guide to the short term start up current that the close rated fuse will tolerate for a few tenths of a second.
230Vac divided by 4.4Aac gives a total primary plus resistor resistance of ~52r
Subtracting 2r for the primary resistance leaves you with a 50r resistor, or 47r
Use a stack of 5off 10r resistors for that, or use a stack of NTCs for that. Choose. I use resistors because they are much cheaper. I bought 1000 5W 10r a few years back.
If your primary is much higher, you may get away with 40r of added resistance.
Fuse each transformer separately.
Use a double pole relay to bypass the two sets of limiting resistors.
You could try using 25r or 30r and see how many times the T2A manages a restart after discharging the capacitors. That would be a maximum current of three to four times the close rated fuse.
That keeps it permanently cold even if repeated restarting is required.
ONE 500VA 230Vac transformer should be close rated fused @ around 2A, but you may find that T1.6A works for years. Or you may have to go up to T2.5A if your voltage is high and the primary inductance of the transformer is low.
The nominal maximum current will be around 500/230 ~ 2.2Aac . There's that T2A fuse rating.
If I double this number to 4.4A, that has given me a good guide to the short term start up current that the close rated fuse will tolerate for a few tenths of a second.
230Vac divided by 4.4Aac gives a total primary plus resistor resistance of ~52r
Subtracting 2r for the primary resistance leaves you with a 50r resistor, or 47r
Use a stack of 5off 10r resistors for that, or use a stack of NTCs for that. Choose. I use resistors because they are much cheaper. I bought 1000 5W 10r a few years back.
If your primary is much higher, you may get away with 40r of added resistance.
Fuse each transformer separately.
Use a double pole relay to bypass the two sets of limiting resistors.
You could try using 25r or 30r and see how many times the T2A manages a restart after discharging the capacitors. That would be a maximum current of three to four times the close rated fuse.
Last edited:
Hi Guys
As AndrewT has succinctly detailed, soft-starting an amp controls both the PT magnetisation inrush current and the cap charging current. This leads to a bit longer limit time being needed if you want fuses that provide protection to anything other than the mains.
Back in the earlier posts, the circuit with R3 across the cap is a common one. R3 is essential in resetting the circuit after power-down - a point I did not see mentioned yet.
Companies I respect use triacs and microncontrollers to soft-start their large toroids, which is something I find aesthetically displeasing. The discontinuity of thyristor conduction should introduce high-frequency noise directly into the PT - and this after the DC blocker has cleaned up the applied mains voltage! Seems like a contradiction of goals to me. The PT will roll off some of that noise but I still do not like this approach.
The use of series Rs with a relay bypass is pretty reliable. I've used a diodeless mosfet AC current limit circuit to good effect (also as the turn-on circuit),but this requires an auxiliary supply. In this circuit the mosfets dissipate some heat during the limiting and there is no set time constant, so any fault in the amp will invoke the limit effect. I try to stay with discrete circuits so they can be repaired if required, and use linear techniques since that is what I understand best. Smarter people than I can use the ucontrollers and maybe also know more about what happens with the triac noise through a toroid.
Have fun
As AndrewT has succinctly detailed, soft-starting an amp controls both the PT magnetisation inrush current and the cap charging current. This leads to a bit longer limit time being needed if you want fuses that provide protection to anything other than the mains.
Back in the earlier posts, the circuit with R3 across the cap is a common one. R3 is essential in resetting the circuit after power-down - a point I did not see mentioned yet.
Companies I respect use triacs and microncontrollers to soft-start their large toroids, which is something I find aesthetically displeasing. The discontinuity of thyristor conduction should introduce high-frequency noise directly into the PT - and this after the DC blocker has cleaned up the applied mains voltage! Seems like a contradiction of goals to me. The PT will roll off some of that noise but I still do not like this approach.
The use of series Rs with a relay bypass is pretty reliable. I've used a diodeless mosfet AC current limit circuit to good effect (also as the turn-on circuit),but this requires an auxiliary supply. In this circuit the mosfets dissipate some heat during the limiting and there is no set time constant, so any fault in the amp will invoke the limit effect. I try to stay with discrete circuits so they can be repaired if required, and use linear techniques since that is what I understand best. Smarter people than I can use the ucontrollers and maybe also know more about what happens with the triac noise through a toroid.
Have fun
On heavy loads' you can use a triac for soft start and once conducting fully in all 4 quadrants, short it out with a rekay.
I've used series resistors and relays on my two big amps - the first uses a 2 kVA core and thr second a 1 kVA core.
I've used series resistors and relays on my two big amps - the first uses a 2 kVA core and thr second a 1 kVA core.
A turned-on triac is essentially a diode and mains DC blockers are made from...diodes!
Would the triac nonlinearity be worse than the diode?
Seems like it should be possible to use the triac as the DC blocker.
Potentially a very neat solution, use phase control to simultaneously soft start the toroid inductive load, ramp up the capacitors and block DC - all with minimal parts, doesn't need chunky resistors or hot NTCs.
Do we need to?
The junction capacitance of the triac could actually be helpful here.
Need to simulate this.
If needed to switch some capacitance across the triac, like in a DC blocker, then perhaps use MOSFETs to keep it all solid-state?
Best wishes
David
No. The DC blocker is the capacitor, or capacitor bank, that passes the Mains AC current. The diodes are there to prevent a high current from overloading/damaging the capacitors.
It's a very standard technique. Wait until Vin is at max, so Ip is at min (90 degree V I phase lag), then start a timer so you Fire the Triac at say 160 degrees. Repeat on negative half. Do this for say 20-50'cycles gradually decreasing phase angle until full conduction.
This takes care of transformer in-rush and charging the cap bank (ramp up time will be dependent on cap bank size - for high cap values, simply extend appropriately).
For DC blocker duty, I do believe you could use a triac. But, without a bypass cap, you will get notches in the mains waveform = noise. Also have to think about what to do with the primary inductive kickback when you do not use a cap across the diode(thermistor) element. For this reason, I personally would prefer to keep the DC blocker function separate.
This takes care of transformer in-rush and charging the cap bank (ramp up time will be dependent on cap bank size - for high cap values, simply extend appropriately).
For DC blocker duty, I do believe you could use a triac. But, without a bypass cap, you will get notches in the mains waveform = noise. Also have to think about what to do with the primary inductive kickback when you do not use a cap across the diode(thermistor) element. For this reason, I personally would prefer to keep the DC blocker function separate.
The transformer and filter capacitance are connected. Yes, there is a short delay before current flows in the secondary but charging of the filter capacitance starts very, very quickly. The two events are inseparable and overlapping unless you have a switch between secondary and filter capacitance.
An NTC can be modelled rather accurately, certainly accurately enough for this work.
Why guess in seconds when a little bit of easy modelling will provide you with greater insight? You can see from the model I provided just how quickly the inrush associated with transformer and filter capacitance occurs (just delete the NTC and run it). Change the transformer model and capacitance amount/form for your actual circuit and observe again. Change the parameters for the NTC to suit the one you have in mind and observe again.
It may take many seconds for the filter capacitance to fully charge, but the soft start limiter (resistor bank or NTC) need not be in place for anywhere near this long. 400ms is a very long time with respect to peak in-rush periods. Want to be a bit more conservative than your updated model suggests? Just round it up another 100ms.
Last edited:
You can model it.
I can't think it and therefore can't model it.
If I can't check the results I will never allow myself to rely on the results.
You stated:
How effective is your model at limiting the rate of charge?
How effective is the added resistance at limiting the primary current such that it does not blow the close rated fuse?
How good is your model of the transformer?
Can you check it?
Now compare how effective a secondary NTC is at limiting the rate of charge of the smoothing capacitors?
I can't think it and therefore can't model it.
If I can't check the results I will never allow myself to rely on the results.
You stated:
Now try modelling the charging rate of the smoothing capacitor bank while you change the primary NTC.
How effective is your model at limiting the rate of charge?
How effective is the added resistance at limiting the primary current such that it does not blow the close rated fuse?
How good is your model of the transformer?
Can you check it?
Now compare how effective a secondary NTC is at limiting the rate of charge of the smoothing capacitors?
Last edited:
This is precisely the purpose of an NTC model in this application. It is a dynamic resistance model. (You can also model its temperature.) Change the parameters of the NTC model to match a different NTC and the profile of the capacitance charging will change accordingly.
You can always test the circuit.
how can you check the transformer model?
Now try comparing different NTC in primary and see how effective it is in limiting the rate of charge of the secondary located smoothing capacitor/s.
Now try comparing different NTC in primary and see how effective it is in limiting the rate of charge of the secondary located smoothing capacitor/s.
A lot of what we have been discussing (including how to use LTspice files) was covered in the thread I linked to in post 31. There was some earlier discussion in the DIYaudio soft start board thread before it was split off into that separate thread. (BTW the thread also included discussion of creapage and clearance which Gudmund raised a few posts back.) The LTspice Yahoo group is a great resource for questions regarding modelling in LTspice.
EDIT: I'm not saying LTspice is the 'be all and end all' and you are certainly much more experienced in EE than I am. But I am saying that it is a useful resource which can assist a lot of desktop thinking and design prior to putting together a circuit for testing.
EDIT: I'm not saying LTspice is the 'be all and end all' and you are certainly much more experienced in EE than I am. But I am saying that it is a useful resource which can assist a lot of desktop thinking and design prior to putting together a circuit for testing.
Last edited:
Regarding NTC in Spice, have a look here and have a look at Helmut Sennewald's notes and sub circuit model in the NTC softstart models I posted and you have downloaded. The latter is in a box to the right of the schematic. Blue material constitutes comments; black is the spice directive.
I will let you explore transformer modelling in LTspice on your own but here is an introduction from Mike Englehardt, LTspice's author. Mark Johnson discussed measuring transformer inductance in the previous thread.
PS: of course one also needs to make sure that the relay model is sensible as well. Note the parameter Coilres in my model and the other parameters in relay.lib (and commented in blue in the box to the top right of the schematic)
I will let you explore transformer modelling in LTspice on your own but here is an introduction from Mike Englehardt, LTspice's author. Mark Johnson discussed measuring transformer inductance in the previous thread.
PS: of course one also needs to make sure that the relay model is sensible as well. Note the parameter Coilres in my model and the other parameters in relay.lib (and commented in blue in the box to the top right of the schematic)
Last edited:
- Status
- Not open for further replies.