No, there are no mounting holes. I can look at adding them if there is room. Are there any other PCB changes before I send the Gerber files to OshPark?
There will be a generic BOM available. I could also do a shopping cart on Digikey; I didn't think Mouser would be a problem. Also, the LED display is not stocked by Digikey or others. I chose the particular LED display for a number of reasons, low cost being one of them (I looked at quite a number of them before deciding on the current part).
The tach can always be used stand-alone (there is no feedback mechanism with this controller and it lacks the necessary resolution (35uHz) needed to implement one).
I've had one other user contact me about stocking the uP. Doing the PCB would be a nice addition since you have to order them in multiples of 3 from OshPark. If you want to do the PCB in qty, let me know, I have board house that charges $150 NRE plus $100 per panel (could probably get 8-16 pcs per panel).
I'm on the verge of releasing all of this to the public domain (except of course the firmware on the uP which I will still control). Does it make sense to start a new thread for ordering the PCB/Parts/uP?
There will be a generic BOM available. I could also do a shopping cart on Digikey; I didn't think Mouser would be a problem. Also, the LED display is not stocked by Digikey or others. I chose the particular LED display for a number of reasons, low cost being one of them (I looked at quite a number of them before deciding on the current part).
The tach can always be used stand-alone (there is no feedback mechanism with this controller and it lacks the necessary resolution (35uHz) needed to implement one).
I've had one other user contact me about stocking the uP. Doing the PCB would be a nice addition since you have to order them in multiples of 3 from OshPark. If you want to do the PCB in qty, let me know, I have board house that charges $150 NRE plus $100 per panel (could probably get 8-16 pcs per panel).
I'm on the verge of releasing all of this to the public domain (except of course the firmware on the uP which I will still control). Does it make sense to start a new thread for ordering the PCB/Parts/uP?
A surprising result from the FET's. They hold from 1R to 2R with just the charge on the gate capacitor for 12 hours. I would have never guessed a cheap capacitor ( Forever brand , Rapid Electonics ) could be that reasonable. This suggests the charging resistor can be whatever value you like. My circa CR= 10 seconds seems a good starting point. The 2R point is verified as 4.82V for this set of FET ( 6 seconds in ). As said before it is not very important where this happens as long as enough time is allowed for a soft start. The window that resembles an ideal NTC device is about 0.5 seconds with 20K. I suspect if 39K used it would be nearly identical to the SL15 2R509 I measured at 2.4 amps and 1 second. The difference being I can have a near zero loss. One can call it a 200R device if doubling the time window to 2 seconds at 39K. No NTC will do that.
If the FET's used are low gate voltage types for 5V logic this should still work. Look to be sure 10V is OK for maximum gate voltage ( should be ). All that might happen is the timing starts earlier.
As the gate charge holds so well I suspect a battery would work for a long time.
I have enjoyed this challenge. Wish my real work was so easy. I hope someone will try it also. Whatever you do don't reject the battery PSU. The complications of PSU ripple are welll worth keeping away from. Although when at near 0R ( R drain-drain ) I doubt it matters. If needing a spare winding on a toroid transformer it can be a metre or two of wire on the toroid upper layer. It's so easy to do and costs a penny or two. Measure lets say 5 turns then calculate what you need.
Keeping R drain-source low has the extra advantage as Hans said of keeping the reverse flow protection diode switched off ( Body diode I think is the name made as a transistor, the zener symbol is the device limit I think and not that it is a true zener that should be exploited ). As he pointed out this allows current to flow in the reverse direction up to just under 0.5 Vrms ( could be 10 amps for some devices ). If so distortion is as a resistor. As far as I can see the source to drain route is OK up to the usual current limits. This body diode is free of charge bonus that needs suplimenting in many cases.
If the FET's used are low gate voltage types for 5V logic this should still work. Look to be sure 10V is OK for maximum gate voltage ( should be ). All that might happen is the timing starts earlier.
As the gate charge holds so well I suspect a battery would work for a long time.
I have enjoyed this challenge. Wish my real work was so easy. I hope someone will try it also. Whatever you do don't reject the battery PSU. The complications of PSU ripple are welll worth keeping away from. Although when at near 0R ( R drain-drain ) I doubt it matters. If needing a spare winding on a toroid transformer it can be a metre or two of wire on the toroid upper layer. It's so easy to do and costs a penny or two. Measure lets say 5 turns then calculate what you need.
Keeping R drain-source low has the extra advantage as Hans said of keeping the reverse flow protection diode switched off ( Body diode I think is the name made as a transistor, the zener symbol is the device limit I think and not that it is a true zener that should be exploited ). As he pointed out this allows current to flow in the reverse direction up to just under 0.5 Vrms ( could be 10 amps for some devices ). If so distortion is as a resistor. As far as I can see the source to drain route is OK up to the usual current limits. This body diode is free of charge bonus that needs suplimenting in many cases.
Pyramid I'm on the verge of releasing all of this to the public domain (except of course the firmware on the uP which I will still control). Does it make sense to start a new thread for ordering the PCB/Parts/uP?[/QUOTE said:
New thread started here: [url]http://www.diyaudio.com/forums/analogue-source/298018-diy-4-phase-sinewave-generator-turntable-motor-drive.html#post4857310[/URL]
I redid the PCB layout and added 4 mounting holes in the corners. The board had to increase in length by 0.25" to accommodate this. The holes are 0.110" diameter, suitable for 4-40 mounting posts.
Rif- I listed you as the source for the pre-programmed chips; how do you want to do the contact info (PM, direct e-mail, website, other?) If you need help ordering the chips, let me know. Have them shipped direct to me, I will program them and ship them to you. You should have my direct e-mail address, please do all correspondence through that channel. Thanks.
I redid the PCB layout and added 4 mounting holes in the corners. The board had to increase in length by 0.25" to accommodate this. The holes are 0.110" diameter, suitable for 4-40 mounting posts.
Rif- I listed you as the source for the pre-programmed chips; how do you want to do the contact info (PM, direct e-mail, website, other?) If you need help ordering the chips, let me know. Have them shipped direct to me, I will program them and ship them to you. You should have my direct e-mail address, please do all correspondence through that channel. Thanks.
Last edited:
Hi Guys
Thought I would share latest results. I put the step up transformers in series on both sides for a 24 to 230v step up. Now the thermistors work. Quess the higher inductance is better for the class d amps???
Here are some measured results of the ntc 5d-20.
Cold 75 degrees F 5.2 ohms
Hot. 3.1 amps. .31 ohms
It takes a about 2 minutes to stabilize. The cool down time is probably the biggest issue here. When you turn it off it takes 53 seconds to get back to 1 ohm/ 2 min 30 seconds to get to 2 ohms / 4 min 30 sec to 3 ohms and 9 min to get to 4 ohms.
I tried start up at 1 ohm level and it seemed to work but 53 seconds is hardly ideal. Well the timed relays are on the way at 5.49 a piece worth the experiment.
Well going to cut .005 off the pulley to dial in closer the 33.3 at 60 hz then finish off with the .01 hz adjustment. Then sit down and enjoy some live albums. Who is first Tom Waits/Jean-Luc Ponty/Grover Washington JR.
Decisions decisions????
Enjoy the ride
Tom
Thought I would share latest results. I put the step up transformers in series on both sides for a 24 to 230v step up. Now the thermistors work. Quess the higher inductance is better for the class d amps???
Here are some measured results of the ntc 5d-20.
Cold 75 degrees F 5.2 ohms
Hot. 3.1 amps. .31 ohms
It takes a about 2 minutes to stabilize. The cool down time is probably the biggest issue here. When you turn it off it takes 53 seconds to get back to 1 ohm/ 2 min 30 seconds to get to 2 ohms / 4 min 30 sec to 3 ohms and 9 min to get to 4 ohms.
I tried start up at 1 ohm level and it seemed to work but 53 seconds is hardly ideal. Well the timed relays are on the way at 5.49 a piece worth the experiment.
Well going to cut .005 off the pulley to dial in closer the 33.3 at 60 hz then finish off with the .01 hz adjustment. Then sit down and enjoy some live albums. Who is first Tom Waits/Jean-Luc Ponty/Grover Washington JR.
Decisions decisions????
Enjoy the ride
Tom
Hi all
Ive started an offline discussion with Pyramid about the uP. Ill keep everyone in the loop.
Im in the US. I have little international shipping experience, but i see that a few of you are in the UK. It's probably much more cost-effective to send several parts to one person for further distribution there?
Ive started an offline discussion with Pyramid about the uP. Ill keep everyone in the loop.
Im in the US. I have little international shipping experience, but i see that a few of you are in the UK. It's probably much more cost-effective to send several parts to one person for further distribution there?
Hans, I have built your post # 238 splitter, and I have a couple of questions about using it. First, what two phase angles do I use for the generator? Does it mater which phase angle goes from the generator to the phase one, and phase two inputs of the splitter? I also have a question about DC offset. You say to use DC offset equal to the bias. The bias is 12 volts, and the most DC offset I can get is 10 volts from my 2 phase generator, do I need to set it at zero and go with input caps in series between the generator and splitter? I assume we are talking a positive DC offset, and not negative. Thanks for all your help Hans, really nice of you.
twystd
twystd
When interchanging the two incoming phases, the effect will be that the motor will reverse direction, that's all.
As long as the "bottom" of the sinewave stays at least 2 Volt above ground, everything will be O.K.
So having for example a 0.4V rms from your generator, 1.13 V Pk-Pk voltage will be generated.
DC voltage should now be at or above 2V + 1.13/2V = + 2.6 Volt
Hans
Hi Hans
Going to experiment with different pulleys and belt/string types. Not sure how .444 diameter is grabbing the silk cord. So lowering the hz I can make a larger diameter pulley. So my question is if I drop the 60 hz motor to 50 hz and the voltage so the motor will not saturate, will it change the phase angle on your splitter. And if it does what is the formula/table to use???
As always Thank You
Tom
Going to experiment with different pulleys and belt/string types. Not sure how .444 diameter is grabbing the silk cord. So lowering the hz I can make a larger diameter pulley. So my question is if I drop the 60 hz motor to 50 hz and the voltage so the motor will not saturate, will it change the phase angle on your splitter. And if it does what is the formula/table to use???
As always Thank You
Tom
Hans, anybody, What 2 phase angles should I use from my generator to the splitter, 0, and 90 degrees?
twystd
twystd
Hi there,
Yes, 0 and 90 as inputs will generate the 3 phases, 0,120, and 240 you need for a 3phase, eg Papst, motor
Yes, 0 and 90 as inputs will generate the 3 phases, 0,120, and 240 you need for a 3phase, eg Papst, motor
Both RC times have to be enlarged by 60/50 when going from 60 to 50 Hz.
So instead of 2 times 13K92 go to 16K7.
Hans
Just completed my MOS FET experiment with a rather good outcome. The circuit is as No351. The problem I had with the NTC resistance is if I made a mistake a brown out makes the amplifier module very unhappy ( at least 7.5 amps ). Often it goes into shuntdown mode which needs a reboot. I doubt that will be something it will tollerate for long. This was only driving a lamp. Today I am driving a very large motor that settles to 1.2 amps. I would estimate the initial current to be > 20 amps. This is for my Garrard 401 turntable that is single phase. I deliberately chose an amplifier at it's limit for these tests ( 100 Watts 2R ). My existing PA inspired design ( C - Audio or H-H 1200 ) never minded the load.
The really great news is how beautifully the dual FET's work. Vastly better than hoped for. Reset seems 100 % reliable ( shorting action of a wafer switch should be OK to get a reset ). 8 seconds to start and about 3 seconds of power up ( my graph suggests less, reality is better as the motor will start sooner than I guessed ). Current peaks at 1.5 amps. I will re-run with cheaper FET's next. Rod Elliot of ESP Audio warns that FET mismatch could be a problem. Maybe. To buy 10 x IRF540 and match them would not be hard. To be honest I didn't see that problem.
Personally I wouldn't use a NTC as too many times I caught it out. I like a relay and 2R resistor. The NE555 resets as easily as my FET's. What I prefer when FET's is a gradual turn on like a cold NTC device. The waveforms seem unchanged which like everyone here is something I believe very important. If it interests anyone the motor I am using is 1.38 Amps 33 1/3 and 1.1 Amps 45. As best I can I use the same current for both meaning I raise the output voltage to suit. It is just possible inside the motor spec to do that. I also run +/- 15 % frequency for 78's. I have two turntables in mind here. The Garrard 601 has no speed change pulley so must be by frequency. The motor I am using is from a 601. It's a big one.
I would say the motor is " almost " too much for this amplifier. The FET's allow it to settle quickly and the current limiting glitches no longer seen on the scope. I did design from the data sheet of the module. I knew it wouldn't be easy. I hope Hans will get what we English say " Making a silk purse from a sows ear ". Whilst it is meant as an put down, real engineering always tries to do this.
The really great news is how beautifully the dual FET's work. Vastly better than hoped for. Reset seems 100 % reliable ( shorting action of a wafer switch should be OK to get a reset ). 8 seconds to start and about 3 seconds of power up ( my graph suggests less, reality is better as the motor will start sooner than I guessed ). Current peaks at 1.5 amps. I will re-run with cheaper FET's next. Rod Elliot of ESP Audio warns that FET mismatch could be a problem. Maybe. To buy 10 x IRF540 and match them would not be hard. To be honest I didn't see that problem.
Personally I wouldn't use a NTC as too many times I caught it out. I like a relay and 2R resistor. The NE555 resets as easily as my FET's. What I prefer when FET's is a gradual turn on like a cold NTC device. The waveforms seem unchanged which like everyone here is something I believe very important. If it interests anyone the motor I am using is 1.38 Amps 33 1/3 and 1.1 Amps 45. As best I can I use the same current for both meaning I raise the output voltage to suit. It is just possible inside the motor spec to do that. I also run +/- 15 % frequency for 78's. I have two turntables in mind here. The Garrard 601 has no speed change pulley so must be by frequency. The motor I am using is from a 601. It's a big one.
I would say the motor is " almost " too much for this amplifier. The FET's allow it to settle quickly and the current limiting glitches no longer seen on the scope. I did design from the data sheet of the module. I knew it wouldn't be easy. I hope Hans will get what we English say " Making a silk purse from a sows ear ". Whilst it is meant as an put down, real engineering always tries to do this.
Just re-run with IRF640 that I have. Works equally well. Comes on slightly sooner indicating a lower gate voltage. Random samples show 4% mismatch ( previous < 0.5%, also at random ). A small amount of heatsinking reqired, the chassis would suit as long as insulated. Some FET come in T0220FP. As far as I know that's ready insulated.
Any progress on Pyramid's dedicated 4 phase signal generator? I screwed up the Hans post 238 splitter. I spent so many years working upside down and mirror image with negatives in the printing business that I developed a sort of dyslexia. The soldering is so tight, and I didn't give enough extra leads to correct it. I'd just have to start over, or better yet, go for the dedicated generator.
twystd
twystd
HI there
I've ordered the boards and chips, just waiting for them to arrive. I've also ordered 1 kit of parts from Mouser but the delivery costs and duty make it an expensive option in the UK. I will source the parts locally for any subsequent builds.
I'm in the process of building a bigger power amp/transformers setup using 60VA toroidals.
As soon as I have all the boards/chips I will get back to the people who have contacted me with finalised prices.
I've ordered the boards and chips, just waiting for them to arrive. I've also ordered 1 kit of parts from Mouser but the delivery costs and duty make it an expensive option in the UK. I will source the parts locally for any subsequent builds.
I'm in the process of building a bigger power amp/transformers setup using 60VA toroidals.
As soon as I have all the boards/chips I will get back to the people who have contacted me with finalised prices.
The FET soft start has run all night at 3 amps using two small postage stamp sized pieces of aluminim as cooling. The IRF640 is not really ideal at that current. IRF530 is cheaper and better. Some 530's are 0.06R, there is an insulated version ( the 540 looks ideal ). The loss of the 640 is about 0.5V per device at 3 amps. That is in the region where the protecion diode comes alive. I saw no wave distortion even so ( on scope, not analyser ). Being conservative the IRF530 should reduce the loss to 0.3V rms per device at 3 amps. I might still add a relay to short them out on my PSU. Hopefully somone needs this idea. I certainly do.
It is so nice to see the current meter take 3 seconds to reach working current and see it peak at only +50% at 1.5 seconds ( 22K charging resistor and 470 uF ). If Rod Elliot is to be believed 50 amps for a toroid is not impossible. That was for 100 volt line added to conventional amps. Much the same. He found the transfomer, open circuit the worse. That makes sense. E&I style transformers are better. Some even have a reluctance gap to help, E&I types do even if you don't want it. If so the output seldom is below 50 Hz when 100 Volt line, the amp is filtered to make sure. It is possible to deconstruct an E&I type and rebuild it with all the E's one end and the I's the other. Sometimes it will do what you want. Push it together against a gasket of printer paper to get about the right air gap. If so use it at 20% original spec. The FET device is easier I would say. The problem with lossy transformers is they often are very noisy. Even $300 ones from Sowter. If so pot them for this use, The 1131g type potting by Robnor is cheap and works well. When toroids I use 4 inch drain pipe. 1 pack will do 3 transformers. If anything they prefer the potting as regards cooling when a 4 inch diameter surface. I use cardboard and Bostik glue to seal the bottoms. More ideal are used pill pots from a pharmacy if they will let you have them.
IRF5305PBF | Infineon IRF5305PBF P-channel MOSFET Transistor, 31 A, 55 V, 3-Pin TO-220AB | Infineon
IRLI540NPBF INFINEON, MOSFET Transistor, N Channel, 20 A, 100 V, 44 mohm, 10 V, 2 V | Farnell element14
If the spec of this one is correct the loss is almost zero and is very cheap. I should be OK without a heat sink. STP75NF75 Mosfet T0220 80A 75V
https://www.rapidonline.com/st-stp75nf75-mosfet-t0220-80a-75v-47-5622
It is so nice to see the current meter take 3 seconds to reach working current and see it peak at only +50% at 1.5 seconds ( 22K charging resistor and 470 uF ). If Rod Elliot is to be believed 50 amps for a toroid is not impossible. That was for 100 volt line added to conventional amps. Much the same. He found the transfomer, open circuit the worse. That makes sense. E&I style transformers are better. Some even have a reluctance gap to help, E&I types do even if you don't want it. If so the output seldom is below 50 Hz when 100 Volt line, the amp is filtered to make sure. It is possible to deconstruct an E&I type and rebuild it with all the E's one end and the I's the other. Sometimes it will do what you want. Push it together against a gasket of printer paper to get about the right air gap. If so use it at 20% original spec. The FET device is easier I would say. The problem with lossy transformers is they often are very noisy. Even $300 ones from Sowter. If so pot them for this use, The 1131g type potting by Robnor is cheap and works well. When toroids I use 4 inch drain pipe. 1 pack will do 3 transformers. If anything they prefer the potting as regards cooling when a 4 inch diameter surface. I use cardboard and Bostik glue to seal the bottoms. More ideal are used pill pots from a pharmacy if they will let you have them.
IRF5305PBF | Infineon IRF5305PBF P-channel MOSFET Transistor, 31 A, 55 V, 3-Pin TO-220AB | Infineon
IRLI540NPBF INFINEON, MOSFET Transistor, N Channel, 20 A, 100 V, 44 mohm, 10 V, 2 V | Farnell element14
If the spec of this one is correct the loss is almost zero and is very cheap. I should be OK without a heat sink. STP75NF75 Mosfet T0220 80A 75V
https://www.rapidonline.com/st-stp75nf75-mosfet-t0220-80a-75v-47-5622
Last edited:
Just hooked up the toroids with a scope to the output ( 130 to 190 V rms ). Interesting to note that the toroid causes start up distortion whilst the E&I didn't. Nothing the amplifier couldn't cope with, no nasty top of the wave oscillation that I have seen on this amplifier ( it looked certain to fail at one point, just switched off in time ). The nice surprise was current peaking was similar via FET soft start as when the E&I type. Big surprise is this 100VA toroid seems to loose far less power. At a rough estimate 25 % better. The E&I being minimum size for the job. If wondering this is parts I have and not selected for task. By chance all the questions answered that I might have avoided if buying in special parts and not wanting to risk a marginal device.
As I am getting a very nice result I will retest with a low loss FET. It might be a mistake. The high loss might be helping stability. That will be a in a weeks time.
I can simulate that using a wire shorting link to the FET's ( IRF640 ).
Now to paint the house, yuk.
As I am getting a very nice result I will retest with a low loss FET. It might be a mistake. The high loss might be helping stability. That will be a in a weeks time.
I can simulate that using a wire shorting link to the FET's ( IRF640 ).
Now to paint the house, yuk.
Following on from print working forcing dyslexia, I doubt that. It might not help if you have it. Many engineers are dyslexic ( I sometimes doubt the ones that aren't ). Dyslexic engineers seem to have the ability to rotate objects inside their heads rather than words ( Alan Blumlein, George Stephenson ). I would say a lucky trade to do that. A similar trait is being left handed. I suspect nature makes a few to do the other tasks. Bob who I interviewed for his job can not read nor write above the age of about 10 ( he didn't say, it became clear over time ). However he can make nearly anything, impossible things also ( truely ). He got his job when he said BSA Gold Star and how he looked after it. About 15 years into knowing Bob he says he hooked up 9 Concorde flight decks. When I said how did he do the paperwork Bob said his work mates did it. The English boss didn't know, the French engineers did! I am dyslexic and wouldn't swap for the world.
On a scale of 1 to 10 all are deaf, blind, dyslexic, autistic. The latter is a great engineering bonus if the type Einstein suffered from. Women hide autism better. I suspect autism is very common and can be a gift.
I have just conected a transformer of slightly better ratio that is 50VA. That's half the size of before. It's almost even in power use, it should be less. That's a surprise as the transformer is large enough. The move to 60VA for this project seems worthwhile. 100 VA might be better.
On a scale of 1 to 10 all are deaf, blind, dyslexic, autistic. The latter is a great engineering bonus if the type Einstein suffered from. Women hide autism better. I suspect autism is very common and can be a gift.
I have just conected a transformer of slightly better ratio that is 50VA. That's half the size of before. It's almost even in power use, it should be less. That's a surprise as the transformer is large enough. The move to 60VA for this project seems worthwhile. 100 VA might be better.