The lamp was not a good idea after all because it works in the wrong direction. When cold resistance is low, when starting to glow resistance increases, it is a PTC so to say.
What you need is an NTC with a resistance that is high at start up, and that because of heating up lowers its resistance.
Epcos has quite a range of NTC's in the lower ohmic range up to 25 Ohm for several dissipation factors when looking on Farnell's website . It is a matter of trial and error.
Hans
Many lamps have a warm resistance about right to do this job. These lamps can also be placed between power amplifier and power supply rails ( two lamps if +/- supply, 2 per amplifier ). An extra capacitor can be added at the amplifier power in terminals to get a small advantage in PSU ripple ( 1000 uF ? forms a Pi filter ). It should also reduce any stability problems of the amplifer due to this resistance. This will also reduce current due to fault conditions. The big advantage of placing the lamp between the PSU and amplifier is it will not affect the damping factor of the amplifier. My conjecture is this is part of what we are doing here, using the damping factor. To a small extent the motor can be controled by the amplifer very much not liking what the load is doing. If a resistor or thermistor ( lamp ) placed in the amplifier output the damping will be reduced. If you like the motor no longer can be seen by the amplifer. Although unlikely the amplifer might have a sweet spot with a small amount of resistance between it and motor. I speculate 1R about right. Take the motor to same motor voltage and hold the motor. If it is quieter the sweet spot exists, hands often can tell ( in the turntable you will not tell so easilly ). My other conjecture is only the 3 rd and 5 th harmonics could be involved as the motor is a filter in it's own right.
These lamps are used instead of thermistors. For example a 20 watt 12 V MR11 lamp is 0.9 Ohms cold. It's hot resistance is 7.2 Ohms ( just measured now and calculated ). I suspect this one would settle at 3 Ohms if doing an ideal job. If possible choose a lamp that could cope with full voltage. One can even put a lamp in the mains input to the PSU. Always check the voltage at the amplifer is high enough to work when running. Switchmode power supplies often will not allow the input to be current limited. They compensate and often switch off. I suspect old style car lamps might have better resistance ranges ( lower voltage side ). I could imagine if using conventional PSU with transformer a lamp into the 115/230 V side could work well. The 60 watt type with filement comes to mind ( 115 or 230 type ). It should flash about 1/2 as bright as full voltage on charging the transformer iron core. Similar on amplifiers changing speed. After that dull red. In the 1930's railways used this type of device as a regulator. Osram made them with a double iron filements. The single lamp at 115/230 V is my favourite.
One way to make an oscillator is to use lamp resistance as a regulator. Although I have seen this for years I only recently tried it. One little lamp I tried zooms form 16 ohms to 160 ohms just on the small current of my meter. Knowing how critical Wien oscillators are it is remarkable how well this 1930's idea works ( Hewlett Parkard ). Even the humble NE5534 can go to better than - 80dB using this idea and a small hand full of extra parts. The lamp I used was Maplin BT44X.
These lamps are used instead of thermistors. For example a 20 watt 12 V MR11 lamp is 0.9 Ohms cold. It's hot resistance is 7.2 Ohms ( just measured now and calculated ). I suspect this one would settle at 3 Ohms if doing an ideal job. If possible choose a lamp that could cope with full voltage. One can even put a lamp in the mains input to the PSU. Always check the voltage at the amplifer is high enough to work when running. Switchmode power supplies often will not allow the input to be current limited. They compensate and often switch off. I suspect old style car lamps might have better resistance ranges ( lower voltage side ). I could imagine if using conventional PSU with transformer a lamp into the 115/230 V side could work well. The 60 watt type with filement comes to mind ( 115 or 230 type ). It should flash about 1/2 as bright as full voltage on charging the transformer iron core. Similar on amplifiers changing speed. After that dull red. In the 1930's railways used this type of device as a regulator. Osram made them with a double iron filements. The single lamp at 115/230 V is my favourite.
One way to make an oscillator is to use lamp resistance as a regulator. Although I have seen this for years I only recently tried it. One little lamp I tried zooms form 16 ohms to 160 ohms just on the small current of my meter. Knowing how critical Wien oscillators are it is remarkable how well this 1930's idea works ( Hewlett Parkard ). Even the humble NE5534 can go to better than - 80dB using this idea and a small hand full of extra parts. The lamp I used was Maplin BT44X.
To get an idea of how well a lamp works.
50 watt 230V GU10 lamp.
Cold 79 ohms.
If a linear resistance 670 watts at 230 V.
Actual resistance 1058 ohms at 230 V. ( 1: 21.6 change ).
At my house I have 246 VAC. This implies 57.2 watts. In fact more like 53 watts as the resistance contiues to rise.
To find a lamp that works should not be too difficult. 60 to 100 watts at the transformer input looks possible. 2 in paralell if too much voltage lost. The warm resistance will vary type of lamp to an other type ( ES to GU ) . Also make to make.
50 watt 230V GU10 lamp.
Cold 79 ohms.
If a linear resistance 670 watts at 230 V.
Actual resistance 1058 ohms at 230 V. ( 1: 21.6 change ).
At my house I have 246 VAC. This implies 57.2 watts. In fact more like 53 watts as the resistance contiues to rise.
To find a lamp that works should not be too difficult. 60 to 100 watts at the transformer input looks possible. 2 in paralell if too much voltage lost. The warm resistance will vary type of lamp to an other type ( ES to GU ) . Also make to make.
Sorry Nigel but I fully disagree.
The problem as described is that the inrush current is causing the problem.
So there are several thinkable solutions
1) using NTC's with a high resistance at start up, something like 8 Ohm shrinking to 1 or 2 Ohm when heating up, exactly the opposite as what a lamp can offer.This is the easiest solution and the first to try.
2) use a zero crossing turn on circuitry
3) let the sinewave come in gradually from zero to 100%, thereby preventing high inrush currents.
1) means the use of 3 NTC's, their value be assessed by trial and error.
2) This needs a bit of thinking, since you do not want to place triacs in the 110 Volt supply, because this will impact negatively on the distortion,
so it has to be in the primary side of the circuitry. Not that easy to realise.
3) this is what seems possible by using a logarithmic ladder network with relays giving a gradual step by step increase of the signal. The question is how many steps, 2, 3 or 4 ?
For each channel you will have to replicate this. With 4 steps you will need 2 small relays per channel, adding up to quite some hardware for 3 channels.
To see if this last approach is feasable at all, you could drive one phase of the papst motor and turn the amplitude of the signal from almost zero to 100% and see what the Main Amp does. If he keeps transferring the signal, the gradual turn on is a safe way to succes,
But it could also be that the Amp starts oscillating because of the inductive load.
in that case a Zobel network, 10 Ohm in series with 100 nF placed at the output of the main amp could help.
I have nothing to try so I cannot do these experiments at my place.
Hans
The problem as described is that the inrush current is causing the problem.
So there are several thinkable solutions
1) using NTC's with a high resistance at start up, something like 8 Ohm shrinking to 1 or 2 Ohm when heating up, exactly the opposite as what a lamp can offer.This is the easiest solution and the first to try.
2) use a zero crossing turn on circuitry
3) let the sinewave come in gradually from zero to 100%, thereby preventing high inrush currents.
1) means the use of 3 NTC's, their value be assessed by trial and error.
2) This needs a bit of thinking, since you do not want to place triacs in the 110 Volt supply, because this will impact negatively on the distortion,
so it has to be in the primary side of the circuitry. Not that easy to realise.
3) this is what seems possible by using a logarithmic ladder network with relays giving a gradual step by step increase of the signal. The question is how many steps, 2, 3 or 4 ?
For each channel you will have to replicate this. With 4 steps you will need 2 small relays per channel, adding up to quite some hardware for 3 channels.
To see if this last approach is feasable at all, you could drive one phase of the papst motor and turn the amplitude of the signal from almost zero to 100% and see what the Main Amp does. If he keeps transferring the signal, the gradual turn on is a safe way to succes,
But it could also be that the Amp starts oscillating because of the inductive load.
in that case a Zobel network, 10 Ohm in series with 100 nF placed at the output of the main amp could help.
I have nothing to try so I cannot do these experiments at my place.
Hans
That's right and it's wrong. The thing to consider is the effect is of heating the lamp is nearly instant. If not a lamp would last 5 mS or less in real life. Usually as the amplifier modules have some tollerence the problem is reduced to a managable size by a linear or non linear series resistance somewhere in the current path. Granted a NTC device is better by a small margin. Remember also the starting needs of the motor should need a kick, is the lamp better? The reason to mention lamps is most of us have some to play with. The main thing to say is something might be needed and where best to put it. I was very surprised how quickly lamps heat. If the amplifier is very sensetive to the inductance usually it is a problem waiting to happen. The old chip L165 liked motors. Most amps don't.
Some power op amps have 1R 100nF Zobels ( TDA series, 2003? ). It should be no problem to use a more drastic Zobel. You will loose some HF, it is of no importance. A single 1R resistor could work in series with the motor. I suspect it could be universally a good thing. It changes the power factor angle a little.
You could use LM317 ( 337 ) as constant current sources in the PSU's. Some LM317 allow up to 2 amps. There is a small voltage loss. I suspect the lamp is better.
I had a similar problem with a 500VDC PSU I made recently. I fitted a resistor in series with the high quality diodes of the soft recovery type. They have a 70 amp peak current rating which looked dangerously close to reality on switch on. The resistor fitted lost 15 VDC. The simple answer was to fit a NE555 timer and relay in a one shot circuit on strip board. The relay comes on in 3 seconds using 2M7 and 1uF polyester, it shorts the resistor. My proudest bit was the power was from a hand wound 10 VDC unregulated supply using about 2 metres of 7 strand insulated wire on the existing toroidal transformer. I did one turn and got a voltage and went from there by measuring the total turn length. 10 minutes work including choosing to try it and about 90% the calculated result. That's where toroids score. I was aiming for 12V and got 10.7VDC. Good enough as 12V relays are fine at 10 VDC. I really need that 15 VDC and now I have it. The circuit seems fail-safe with the small 1uF 63V polyester high grade capacitor. I could not fool it regardless of how I tried. 555 seem OK up to 10M ohms if a high grade capacitor used. The smaller the capacitor the quicker the discharge.
Zero crossing can give unseen problems. It's well worth a try. If you had a zero crossing opto coupler and 2 x N type FET as AC relay you might be OK. Join sources and gates, use drains as wire in and wire out. The FET is coupled with a forward biased diode on each half cycle from the other FET. It looks a recipe for disaster. It is better than I ever guessed down to the 20 Vrms level I measured if 50 Hz ( IRF640 that I had ). The FET's must be fully saturated, if a protection diode not fitted it is doubly needed as it is also the half cycle conductor. 9VDC should work from sources and gates to hard switch them on ( note they are joined and a 9V PP3 battery to test the idea is all you need, drains are free and not joined ). A standard relay is about 4 to 10 mS. That would be useless for zero crossing. I think a switched on FET is about for example 200 R x 3nF ( gate stopper resistance x Cgs x 2 ). To 1CR that is 1.2 uS. I think that is the worst it can be as 62.3 % gate maximum even with high Cgs should mean fully switched on in 1 uS.
I had a 24V Papst motor on a Michell. I remember it was a tough thing to drive. I built it a descrete amplifer good for about 100 watts using big T03 Darlington devices by Motorola. +/- 50 VDC PSU. It was a PNP longtail pair, 8 mA NPN VAS and Darlingtons ( MJ11015/16BG ? ). My Zobel was 100 nF 4R7 3 watt. One thing that might help is to use the highest gain on the power amp to encourage stability. The distortion should be miniscule as it is LF. I am thinking in terms of gain 100 or at least 30. I like L165 as it is OK at inverting gain of 1 driving a motor. Try if possible the inverting input of amps if available. Often it is better when marginal.
Some power op amps have 1R 100nF Zobels ( TDA series, 2003? ). It should be no problem to use a more drastic Zobel. You will loose some HF, it is of no importance. A single 1R resistor could work in series with the motor. I suspect it could be universally a good thing. It changes the power factor angle a little.
You could use LM317 ( 337 ) as constant current sources in the PSU's. Some LM317 allow up to 2 amps. There is a small voltage loss. I suspect the lamp is better.
I had a similar problem with a 500VDC PSU I made recently. I fitted a resistor in series with the high quality diodes of the soft recovery type. They have a 70 amp peak current rating which looked dangerously close to reality on switch on. The resistor fitted lost 15 VDC. The simple answer was to fit a NE555 timer and relay in a one shot circuit on strip board. The relay comes on in 3 seconds using 2M7 and 1uF polyester, it shorts the resistor. My proudest bit was the power was from a hand wound 10 VDC unregulated supply using about 2 metres of 7 strand insulated wire on the existing toroidal transformer. I did one turn and got a voltage and went from there by measuring the total turn length. 10 minutes work including choosing to try it and about 90% the calculated result. That's where toroids score. I was aiming for 12V and got 10.7VDC. Good enough as 12V relays are fine at 10 VDC. I really need that 15 VDC and now I have it. The circuit seems fail-safe with the small 1uF 63V polyester high grade capacitor. I could not fool it regardless of how I tried. 555 seem OK up to 10M ohms if a high grade capacitor used. The smaller the capacitor the quicker the discharge.
Zero crossing can give unseen problems. It's well worth a try. If you had a zero crossing opto coupler and 2 x N type FET as AC relay you might be OK. Join sources and gates, use drains as wire in and wire out. The FET is coupled with a forward biased diode on each half cycle from the other FET. It looks a recipe for disaster. It is better than I ever guessed down to the 20 Vrms level I measured if 50 Hz ( IRF640 that I had ). The FET's must be fully saturated, if a protection diode not fitted it is doubly needed as it is also the half cycle conductor. 9VDC should work from sources and gates to hard switch them on ( note they are joined and a 9V PP3 battery to test the idea is all you need, drains are free and not joined ). A standard relay is about 4 to 10 mS. That would be useless for zero crossing. I think a switched on FET is about for example 200 R x 3nF ( gate stopper resistance x Cgs x 2 ). To 1CR that is 1.2 uS. I think that is the worst it can be as 62.3 % gate maximum even with high Cgs should mean fully switched on in 1 uS.
I had a 24V Papst motor on a Michell. I remember it was a tough thing to drive. I built it a descrete amplifer good for about 100 watts using big T03 Darlington devices by Motorola. +/- 50 VDC PSU. It was a PNP longtail pair, 8 mA NPN VAS and Darlingtons ( MJ11015/16BG ? ). My Zobel was 100 nF 4R7 3 watt. One thing that might help is to use the highest gain on the power amp to encourage stability. The distortion should be miniscule as it is LF. I am thinking in terms of gain 100 or at least 30. I like L165 as it is OK at inverting gain of 1 driving a motor. Try if possible the inverting input of amps if available. Often it is better when marginal.
Though my setup uses a 1 ohm resistor in series with the amp output, currently shorted out by a relay after 3 seconds I'm not very happy with the extra complexity. I've ordered a selection of NTC power thermistors to see if these work.
My Papst motors haven't yet made it into my turntable, but they run quite happily with 1 ohm resistors at startup, so I'm a bit confused why others are having problems. FWIW I've also tried running a much larger Papst capstan motor from a Revox PR99, and though it does take a few seconds to get up to speed there's no problem with the amps or the transformers.
My Papst motors haven't yet made it into my turntable, but they run quite happily with 1 ohm resistors at startup, so I'm a bit confused why others are having problems. FWIW I've also tried running a much larger Papst capstan motor from a Revox PR99, and though it does take a few seconds to get up to speed there's no problem with the amps or the transformers.
Attachments
If you can, find out the ideal resistance. It is highly debatable if 0R is best. The main debate is damping factor good or bad. I fully understand disliking complexity. I have a hunch 12V lamp should be excellent. Perhaps car type H4 using series or paralell filements. Often these are very cheap. MR16 12V in 20 or 50 W also.
Hey Ralph
What kind and how big is your power supply? This 6 wire motor is nice but large. We are guessing that the mean well 24v 150watt 6.3 amp power supply is right on the limit.
For example the motor will start to run at 25 volts slowly then if you bring the signal it will increase rpm until reaches around 50 volts then the power supply drops from 24 volts to the amps to about 12v. Also if you hit the power supply with max signal it does not go as high as if you bring it up gradually. So the conclusion we came to is the protection circuits in the power supply is kicking in never allowing the motor to lock in.
Well have a 350 watt unit on the way, maybe that will help. Like the thermistor idea regardless if it the main problem here. How many amp thermistor did you order and what ohms at cold ?
Thanks Tom
What kind and how big is your power supply? This 6 wire motor is nice but large. We are guessing that the mean well 24v 150watt 6.3 amp power supply is right on the limit.
For example the motor will start to run at 25 volts slowly then if you bring the signal it will increase rpm until reaches around 50 volts then the power supply drops from 24 volts to the amps to about 12v. Also if you hit the power supply with max signal it does not go as high as if you bring it up gradually. So the conclusion we came to is the protection circuits in the power supply is kicking in never allowing the motor to lock in.
Well have a 350 watt unit on the way, maybe that will help. Like the thermistor idea regardless if it the main problem here. How many amp thermistor did you order and what ohms at cold ?
Thanks Tom
Hi Tom
My power supplies (I've tried two) are 24V 5A, and they both provide all the power I need with no dropout.
The NTC thermistors I have coming are all 5 ohm ones, 2A,3A, and 5A ones as I'm not sure what I need. The hot resistance is between 0.15 and 0.25 ohm.
I'm still waiting for my 6 wire Papst (probably stuck in customs
)
I'm hoping my smaller Papst (in the picture) is similar in size to the original one used in the Thorens TD124. it's 64mm (~2.5") across the outside of the rotor.
My power supplies (I've tried two) are 24V 5A, and they both provide all the power I need with no dropout.
The NTC thermistors I have coming are all 5 ohm ones, 2A,3A, and 5A ones as I'm not sure what I need. The hot resistance is between 0.15 and 0.25 ohm.
I'm still waiting for my 6 wire Papst (probably stuck in customs
)I'm hoping my smaller Papst (in the picture) is similar in size to the original one used in the Thorens TD124. it's 64mm (~2.5") across the outside of the rotor.
Hi Ralph
Well good news or bad news depending on your goal. The motor is pretty big. The outer steel band is 3.650"Dia. Black fan looking plastic part 3.219 dia. Bracket height 2.820. Mounting plate 3.387 sq. And the shaft is 8mm dia.
Weighs in at 2lb 12 oz.
I am guessing this will be the big dog on the porch. But will take up more space and eat more food.
Tom
Well good news or bad news depending on your goal. The motor is pretty big. The outer steel band is 3.650"Dia. Black fan looking plastic part 3.219 dia. Bracket height 2.820. Mounting plate 3.387 sq. And the shaft is 8mm dia.
Weighs in at 2lb 12 oz.
I am guessing this will be the big dog on the porch. But will take up more space and eat more food.
Tom
That is a bit of a beast, but in a P*****g contest my revox capstan motor has a rotor outer diameter of 3.5 inches, and weighs just on 4lbs. No wonder it takes a few seconds to get up to speed I suspect however that it isn't a synchronous Papst as the outer rotor doesn't have any visible structure.
I suspect however that it isn't a synchronous Papst as the outer rotor doesn't have any visible structure.Sorry for the off-topic aside yesterday, back to the script:-
I've assembled the following
UDB1302S Dual DDS Source TTL Signal Generator 60MHz Sweep Frequency Counter (thebay)
Pyramid's two -to- three phase converter (but using split supplies) #172
and my original amplifiers, transformers, and Papst motor.
Nothing very clever in this setup, except for finding out the effect of some of the variations possible.
With the DDS generator set to 60.00hz the motor speed is exactly 1800 rpm, and maintains that speed until I load the motor up to the point where it drops out and stops.
Changing the input frequency up by 0.1HZ the speed goes to 1803, and 0.2Hz to 1806 rpm, within the limits of my cheap Chinese tacho. This does at least show that the DDS generator is doing what it says.
More interesting is the fact that changing the phase relationship by even 1 degree makes a noticeable change to the motor's vibration; 90 is the only way to go, at least with my motors.
I've assembled the following
UDB1302S Dual DDS Source TTL Signal Generator 60MHz Sweep Frequency Counter (thebay)
Pyramid's two -to- three phase converter (but using split supplies) #172
and my original amplifiers, transformers, and Papst motor.
Nothing very clever in this setup, except for finding out the effect of some of the variations possible.
With the DDS generator set to 60.00hz the motor speed is exactly 1800 rpm, and maintains that speed until I load the motor up to the point where it drops out and stops.
Changing the input frequency up by 0.1HZ the speed goes to 1803, and 0.2Hz to 1806 rpm, within the limits of my cheap Chinese tacho. This does at least show that the DDS generator is doing what it says.
More interesting is the fact that changing the phase relationship by even 1 degree makes a noticeable change to the motor's vibration; 90 is the only way to go, at least with my motors.
Hi Tom,
I'm using a dual output DDS generator, and setting the outputs 90 degree out of phase, then I'm using a 2 x 90 to 3 x 120 circuit as shown by Pyramid on post #172 to get my three phases.
My reasoning for this is to avoid any analog phase shifting using capacitors as the time constants have to be shifted if you change frequency whereas #172 allows the creation of 3 x 120deg phase signals from 2 x 90 deg at any sensible frequency.
I'm using a dual output DDS generator, and setting the outputs 90 degree out of phase, then I'm using a 2 x 90 to 3 x 120 circuit as shown by Pyramid on post #172 to get my three phases.
My reasoning for this is to avoid any analog phase shifting using capacitors as the time constants have to be shifted if you change frequency whereas #172 allows the creation of 3 x 120deg phase signals from 2 x 90 deg at any sensible frequency.
That is why I advised to use 0.1% resistors in the circuitry generating 120 and 240 degrees and to keep the incoming phase exactly at 90 degrees.
The slightest misalignment not only results in phase errors but also in amplitude errors of the 3 phases.
1% shift in the incoming 90 degrees shifted signals, not only results in almost 1 degree shift in the 120 and 240 signals, but also in 1.5% amplitude difference between the 3 outgoing signals.
Hans
I mentioned a MOS FET relay. Looking at my saved images this is very close to what could be used. It was a bit of fun with parts I had. Although not a super low distortion PSU you can see nothing changes much via a tough load using the FET's. I didn't exspect that due to the low voltage. The low frequency must be helpful no doubt. As the FET's are fully saturated the loss is very low.
Find a way to zero volts switch it and who knows. I looked at using the gate capacitance to work as a delay with exponential curve. 10M is about where you would have to be. No idea if that begins to work. If looks just possible. It would change sample to sample. My guess is not enough to worry. The two gates together helps that as the Cgs is double. The protection diode inside the FET helps the conduction half cycle to half cycle.
Ironically I didn't do this for anything like this. It was to test an ultra high voltage concept without risking the parts cost. It worked. It ran 1KW with ease.
Hi Hans,
I agree totally about the required accuracy of the resistors in the phase splitter; I don't have 0.1%, only 2%, but I have a few hundred of each in the E24 range, enough for me to select individual resistors of the necessary value. (or a combination in the case of the 8.66k). My test meter isn't that accurate, but it is repeatable, and all the resistors are in the same range.
I'm using op37 op-amps as these gave the best results in simulation, as liitle as 5uV p~p error.
In real life I've got an error of about 1~1.5mV (summing the three outputs from the phase splitter) with an output of 750mV from each phase; "good enough for military work".
I will rework the phase splitter using the correct resistors before I button up the project but it is working fine.for test purposes.
My NTC thermistors arrived this morning and I've tried a couple of values; my recommendation so far is the NTC 5D-9, this has a 'cold' resistance of around 5 ohms, and drops 0.5V in use, It does heat up in a second or two, but slowly enough for the amp not to saturate the transformer.
Hi Ralph,
That's a great achievement. Such an easy and low cost solution should solve all problems.
Hans
It still would be a good idea to remove the NTC device with a relay of some sort when the critical time period has passed as it gives higher damping factor. That might have some effect. If it is saturation one could try AC coupling. A high value non polar capacitor could work. Just a thought.
Hi Nigel,
Opposite to your suggestion, the protection diodes never come to conduction.
If they did, harmonic distortion would be very high.
Fact is that the Fets are conduction all the time after switch on and are behaving almost like linear resistors in the mOhm range.
But it is certainly a possible way to switch the motor load, although it seems that the NTC that Ralph inserted is a much easier way to get things under control.
Hans