TMH-260 Turbo

Every fusor and fusion system seems to need a vacuum. This area is for detailed discussion of vacuum systems, materials, gauging, etc. related to fusor or fusion research.
Al Nejati
Posts: 12
Joined: Fri May 07, 2021 11:28 pm
Real name: Al Nejati

Re: TMH-260 Turbo

Post by Al Nejati »

I have several questions but the most important observation is that you may need diodes from each motor phase to 40 V, as in the attached schematic for a unipolar stepper driver.

Without that, you'd be getting inductive voltage spikes every time a mosfet turns off. These spikes can very easily kill the mosfets. The zener/avalanche diodes (D1-D4) probably aren't doing anything to protect the mosfets from these spikes:
1. They probably lack the ability to actually carry the amount of current required here (up to 7-8 A spikes, or 1-2 A continuous)
2. Their breakdown voltage (120v) is higher than what the transistors can tolerate anyway.
Attachments
download.png
Gustavo Cervantes
Posts: 27
Joined: Wed Apr 04, 2018 9:17 pm
Real name: Gustavo
Location: Guadalajara, Mexico

Re: TMH-260 Turbo

Post by Gustavo Cervantes »

Not sure why i wrote IRF530, Im actually using IRF630's,
When i started working on this controller i actually was using Flyback diodes instead of Zeners to disipate the energy, however i noticed that the common wire was heating up, which stopped after switching to the zeners, the ones im using are 1N5381B
On the board i forgot to ad a few resistors and the gate protection zener, i bodged them on this last effort.
Attachments
WhatsApp Image 2021-05-08 at 9.36.57 PM.jpeg
WhatsApp Image 2021-05-08 at 9.36.58 PM.jpeg
Al Nejati
Posts: 12
Joined: Fri May 07, 2021 11:28 pm
Real name: Al Nejati

Re: TMH-260 Turbo

Post by Al Nejati »

COM wire overheating is due to the following reason. Each motor winding acts like a transformer. Whenever a half-winding is on, the other half-winding has opposite voltage across it, and now because there's a diode across it too, the transformer is effectively shorted. Current grows through the motor winding without bound.

The solution to this is to instead have the flyback diodes go from the transistor drain to 2*V_supply (80V), rather than 40 V. It's hard to explain this briefly in words, it's best to see this in action in the circuitjs app: https://tinyurl.com/yg3y5ue7
(The transformer is simulating one of the two motor windings; the 20 ohm resistor is simulating the 'load' on the motor).

The switch is set to '80V', if you instead flip it to 40V you can see the current build-up effect occurring.

If the supply doesn't have 80V output, you can 'simulate' it by having a capacitor with a 80V zener, like follows: https://tinyurl.com/yje82yck

The 100 ohm resistor is just arbitrarily chosen; I actually don't know the right value to use here. It would have to be calculated based on allowable voltage drop and dissipation.
Al Nejati
Posts: 12
Joined: Fri May 07, 2021 11:28 pm
Real name: Al Nejati

Re: TMH-260 Turbo

Post by Al Nejati »

That circuit will both properly snub the voltage spikes and dampen the current spikes through the components.
Gustavo Cervantes
Posts: 27
Joined: Wed Apr 04, 2018 9:17 pm
Real name: Gustavo
Location: Guadalajara, Mexico

Re: TMH-260 Turbo

Post by Gustavo Cervantes »

Well thats seems familiar, thats the topology of a switch mode power supply, to be honest the zener diode idea i took from another guy, seems to work for him, not sure why my implementation is giving me problems.
I will do some measurements and report back.
Thank you Al for your input on this
Al Nejati
Posts: 12
Joined: Fri May 07, 2021 11:28 pm
Real name: Al Nejati

Re: TMH-260 Turbo

Post by Al Nejati »

No problem!

Actually I have a question for you: how did you drive the hall effect sensors to get that scope output?
Gustavo Cervantes
Posts: 27
Joined: Wed Apr 04, 2018 9:17 pm
Real name: Gustavo
Location: Guadalajara, Mexico

Re: TMH-260 Turbo

Post by Gustavo Cervantes »

Since it was like 2 years ago, im not too sure on the details, i just remember probing with a voltmeter to find the windings, once i isolated those i just took the remaining cables and started aplying 5v and ground to a bundle of 3 with a current limited source until i got a signal out.
And its not a scope it's the serial ploter from the arduino IDE, I did not have a scope then so I had to get creative.
Al Nejati
Posts: 12
Joined: Fri May 07, 2021 11:28 pm
Real name: Al Nejati

Re: TMH-260 Turbo

Post by Al Nejati »

Chris Giles wrote: Tue May 22, 2018 4:42 pm Hi Gustavo,

it looks like you figured it out, but in case it's any help note that the pinout of the motor's connector and its internal circuit are shown on p9 of the user manual for the TCP380 controller (you will find PDF via Google). I don't know what those two components with a Theta symbol by them are. Maybe the resistor shown from pin P to N is a thermistor.

I can also see the service manual for the TCP310. It uses IR3525A PWM controller ICs. I wouldn't fancy trying to replicate a controller from scratch but I wonder that a generic motor controller board could be adapted using the knowledge from the manuals. How did you get on?

Regards
Chris
The components with theta symbol are the thermistors (PTC type). The resistor from P to N is a feedback resistor.
Al Nejati
Posts: 12
Joined: Fri May 07, 2021 11:28 pm
Real name: Al Nejati

Re: TMH-260 Turbo

Post by Al Nejati »

Also I think I figured out how to properly drive the hall sensors.

As can be seen in the pdf that Chris posted, the hall sensors are series-connected. They are probably intended to be driven with constant current. Based on the manual for the TPH062, I tried a constant current supply of 20 mA, with F being the negative side and G being the positive side of the current supply. Then, the first hall sensor H1 develops either a positive or negative voltage between L and H, and the second sensor H2 will develop a voltage between J and K.
Gustavo Cervantes
Posts: 27
Joined: Wed Apr 04, 2018 9:17 pm
Real name: Gustavo
Location: Guadalajara, Mexico

Re: TMH-260 Turbo

Post by Gustavo Cervantes »

Ok it seems that I misunderstood your last question, at the moment I'm driving the hall sensors with 5v directly, the the output gets compared by the lm2903 to a fixed voltage set via a 10 turn preset that's how I'm getting a 0 or 1 from them,
that's what I feed to the decoder down the line.
I knew there was something else going on there, to be honest I was wondering how the Hall's were connected, but since this seemed to work well enough for my purposes I decided to carry on as is.
Al Nejati
Posts: 12
Joined: Fri May 07, 2021 11:28 pm
Real name: Al Nejati

Re: TMH-260 Turbo

Post by Al Nejati »

Just an update on how I'm currently going. I've got the first version of the driver sent off for pcb fab. The circuit has current sense (with low-pass filter), constant-current hall drive, and (optional) common-mode bypass for potential noise on the hall sense circuit. Time to see how it works in practice. Schematic attached.
Attachments
Screenshot from 2021-07-25 11-53-52.png
User avatar
Richard Hull
Moderator
Posts: 14991
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: TMH-260 Turbo

Post by Richard Hull »

Hitting the issue the hard way. Good luck with you efforts to spin up your turbo using a self-designed turbo controller. Keep us in the loop on how it turns out, once assembled.

Richard Hull
Progress may have been a good thing once, but it just went on too long. - Yogi Berra
Fusion is the energy of the future....and it always will be
The more complex the idea put forward by the poor amateur, the more likely it will never see embodiment
Al Nejati
Posts: 12
Joined: Fri May 07, 2021 11:28 pm
Real name: Al Nejati

Re: TMH-260 Turbo

Post by Al Nejati »

Thanks Richard.

Some more design notes, in case it's helpful to anyone.

The mosfets are driven through a pair of drivers, each driver controlling the two windings of the same phase. So A and C are controlled by one driver, and B and D are controlled by the other driver. The benefit of doing this vs putting A and B on the same driver is that you get hardware shoot-through prevention and dead time, which is important for soft switching (more on this below).

Ideally, you don't want the mosfet for each phase to be fully on during the whole part of the cycle; instead you want to have PWM control. This enables a more controlled ramp-up of the motor with minimal torque ripple, which helps extend bearing life, and eliminates the need for multiple power supplies for the motor. Because PWM involves switching at relatively high frequencies, it is best to avoid hard switching of the mosfets to reduce switching losses. This turns out to be straightforward to do. Whenever a winding of a phase turns off, the induced back EMF 'swings' the other winding down towards zero. So, for example, when winding A is on, C is at 2xV_supply. When winding A turns off, it swings up towards 2xV_supply, and thus C swings down to near 0. The dead-time allows the other mosfet to swing back to 0V as it turns on, reducing turn-on losses. Turn-off losses for A can be reduced by switching it off as quickly as possible and having some capacitance (in this case, a few hundred pF) between A and ground. The mosfet's own C_ds is sufficient for this here.
Al Nejati
Posts: 12
Joined: Fri May 07, 2021 11:28 pm
Real name: Al Nejati

Re: TMH-260 Turbo

Post by Al Nejati »

Got pcb back and wired everything up, and it works. The pcb has an inexpensive STM32F103 board stuck on top to control everything. Done a few basic tests spinning up to very low speed (~6000 rpm) just to make sure current draw, voltage, etc. are within expected bounds. Will spin up to higher rpms when I put my actual vacuum chamber together, and I add the hardware watchdog. The hardware watchdog is important when using a microcontroller to generate the drive frequency. Picture and short video attached (sorry for poor quality). The rundown time (unpowered) from 6000 rpm to 0 rpm is about a minute or so.

Some notes: There's more electrical noise from the motor drive on the hall sensor lines than I expected. There's likely some kind of coupling inside the motor. This causes a problem because the hall sensor voltage output is trapezoidal and goes near zero for a good period of time during each cycle before polarity reversal. This can cause spurious output from the comparators. I've added some debouncing in software, but this is less than ideal. I'll look at more hardware noise filtering in the next prototype.

Current draw at 6000 rpm is about 0.7 A @ 20 V. The mosfets are cold to the touch. I noticed some noise from the turbo around 4800 rpm, I wonder if this is just a resonance and not to worry about, or if there's a problem with my pump. Is your pump similar, Gustavo?
Attachments
IMG_20210822_114946.jpg
VID_20210822_114957.mp4
(2.63 MiB) Downloaded 264 times
IMG_20210822_115409.jpg
Denis Tikhonov
Posts: 3
Joined: Mon Oct 23, 2023 2:32 am
Real name: Denis Tikhonov
Location: Yerevan

Re: TMH-260 Turbo

Post by Denis Tikhonov »

Hi Gustavo and Al, I'm impressed with your result, could you kindly re-upload the schematics since they all don't seem to open?

I have a Pfeiffer TPH180 with no controller, and the motor coils and hall sensors in it have the same arrangement as in the 260 you have spun up. They also seem to share the same TCP380 controller.

You would help a lot to avoid the extra pain of figuring out the signal and phase sequence.

Many thanks in advance.
Post Reply

Return to “Vacuum Technology (& FAQs)”