Electrical Power Steering Assist design prob

[kinarfi]

This design looks good on paper and on the bench, as long as I don't load the FETs, that is connect the drains. When I get this all tuned and balanced, I get a scope picture like this,

This is TP 9 & 10. TP2 should be a smooth line at 5.00 vdc and it is, but when I connected a resistive load to the Power FET driven by OP7, I got this on the scope, Yellow trace, Green trace is the saw tooth I use to create my pulse width output.(no torque on steering wheel)(and only one FET connected)
When I apply enough torque to the steering wheel to get the Pulse width to 100%, TP2 goes flat and smoother again.
My apologies if this is double posting, but I still have a problem. I'm planning on changing the Wheatstone bridge amp when I get the IC, Maybe that will fix every thing.
Does anyone have any suggestions on understand why TP2 distorts,
Thanks,
Kinarfi

 

[BrownOut]

The opamp feeding TP2 has ALOT of gain. I think it's being driven to the rail. Also make sure your rails are clean when you load the drivers.

 

[BrownOut]

Also please put the component designators on the schematic, so I can references the those in the circuit description.

 

[kinarfi]

The opamp feeding TP2 has ALOT of gain. I think it's being driven to the rail. Also make sure your rails are clean when you load the drivers.

Got any suggestions for correcting this? Correct me if wrong, Rails refers to upper and lower voltage, in this case 10.0 vdc and 0, and clean is well regulated and filtered/capacitor + Load the drivers?
Thank you, & I got the id schematic there now.
Kinarfi

 

[Diver300]

The opamp feeding TP2 has ALOT of gain. I think it's being driven to the rail. Also make sure your rails are clean when you load the drivers.

That opamp is amplifying the strain gauge, so it needs lots of gain. The gain and zero are adjustable, to allow for variations in the strain gauge. TP2 should be at around 5V with no load on the steering wheel, and go up to near 10 V when the steering it pulled one way, and down to near 0 V when it is pulled the other way. That is torque on the steering wheel, not how far it is turned.

What worries me on an application like that is the failure modes, as they could affect control of the car.

 

[kinarfi]

That opamp is amplifying the strain gauge, so it needs lots of gain. The gain and zero are adjustable, to allow for variations in the strain gauge. TP2 should be at around 5V with no load on the steering wheel, and go up to near 10 V when the steering it pulled one way, and down to near 0 V when it is pulled the other way. That is torque on the steering wheel, not how far it is turned.

What worries me on an application like that is the failure modes, as they could affect control of the car.

Correct, The TP2 actually only varies about 2 to 2.5 volts before the torque is enough to turn the wheels manually and with the saw tooth being about 2 volt p-p, that's enough to have the output full on. After you click on the schematic, go to page 2 of the full album and you can see some of the mechanical parts. or click this.

My concern right now is 'why is TP2 oscillating when I hook up the coils and what can I do to stop it'.
If the 10 volt regulator or power fails or is turned off, C1 deenergizes and it's contact opens and the drag from a shorted motor become the drag of mechanics and acts as a steering damper, (practical knowledge, not theory) and the off switch is near by and this is for off road and sand dunes. Thanks for your concern, it is one of mine also.
Thanks
Kinarfi

 

[BrownOut]

I get that, but why does the gain need to be so high? Is the output of a strain gauge really so low that it needs to be amplified by several hundreds? Opamps that have that much gain are prone to break into oscillations.

 

[Nigel Goodwin]

Why reinvent the wheel?, my Fiat Punto has electrically assisted power steering, and has a 'girlie button' which increases the gain of the steering, for easy parking.

 

[kinarfi]

I get that, but why does the gain need to be so high? Is the output of a strain gauge really so low that it needs to be amplified by several hundreds? Opamps that have that much gain are prone to break into oscillations.

Yes, read up on it some, It puts out in the 100 micro volt range, short course in strain gauge: 1 I can't believe they work at all 2 they are superglued to the steel shaft and sense the stress of the steel as strain (torque) is applied by stretching or contracting with the surface of the steel and if I remember correctly, the change in resistance is .7 ohm max, it may have been .7 miliohms, I'm checking.
Anyway, you think it is the high gain that is making things oscillate, why is it stable until I connect loads to the FETs?
Thanks
Kinarfi

 

[kinarfi]

Why reinvent the wheel?, my Fiat Punto has electrically assisted power steering, and has a 'girlie button' which increases the gain of the steering, for easy parking.

I WILL look into that, but I am having fun trying to learn more about OPamps and FETs and gaining knowledge and experience, can you tell me more about your Fiat Punto Model year and etc. to help me search.
Oh yea, Why reinvent the wheel? Your Fiat Punto is a reinvention of the Stone wheel, and it's better than the stone wheel. Do you have any idea that might help with my wheel?
Thanks
Kinarfi

 

[BrownOut]

Yes, short course in strain gauge 1 I can't believe they work at all 2 they are superglued to the steel shaft and sense the stress of the steel as strain (torque) is applied by stretching or contracting with the surface of the steel and if I remember correctly, the change in resistance is .7 ohm max, it may have been .7 miliohms, I'm checking.
Anyway, you think it is the high gain that is making things oscillate, why is it stable until I connect loads to the FETs?
Thanks
Kinarfi

Are you sure that when the opamp isn't oscillating, that it isn't stuck to the power rail? You showed the scope capture for the FET's but not for the opamp. Did you check that the rails are clean?

 

[kinarfi]

Are you sure that when the opamp isn't oscillating, that it isn't stuck to the power rail? You showed the scope capture for the FET's but not for the opamp. Did you check that the rails are clean?

Did you check that the rails are clean? Not sure what you mean by this.

The capture of the FETs was taken at the gate to the FETs and my analysis of it is that it shows that TP2 is stable at 5.00 volt and out puts of OP1 and OP2 are balanced with the valley of OP1 and the peak of OP2 are intersecting TP2 equally. If I apply any torque at all one trace disappears and the other widens with increasing torque until it is a solid line, "railed?"
I guess that would be another clarification question: Are OP 1, 2, 3, & 4 considered to be operating between the rails, output varies, and OP 5, 6, 7, & 8 considered on the rails, either full on or full off?

 

[kinarfi]

The Strain Gage
Strain gauge info in case some one wants to know

 

[Nigel Goodwin]

I WILL look into that, but I am having fun trying to learn more about OPamps and FETs and gaining knowledge and experience, can you tell me more about your Fiat Punto Model year and etc. to help me search.

It's a 2003 Mk2 Punto, the Mk1 Punto's didn't have power steering - I'm presuming the new Mk3 versions use a similar system?.

 

[marcbarker]

I've not studied the circuit closely, but these questions may help.

is it an EMC problem? I think someone mentioned this already (supplies being clean or not).

Isn't the strain guage amplifier meant to be on a different PCB to the switching circuit? Or at least segregated on it's own ground return. Also I don't see any supply rail decoupling on the SG amplier. was a tried and tested PCB used?

How does the circuit perform when tested 'open-loop' ? does the oscillation only occur when closed loop?

 

[kinarfi]

I've not studied the circuit closely, but these questions may help.

is it an EMC problem? I think someone mentioned this already (supplies being clean or not).

Isn't the strain guage amplifier meant to be on a different PCB to the switching circuit? Or at least segregated on it's own ground return. Also I don't see any supply rail decoupling on the SG amplier. was a tried and tested PCB used?

How does the circuit perform when tested 'open-loop' ? does the oscillation only occur when closed loop?

is it an EMC problem --Define please, to make sure we speak the same language
this strain gauge is just 1/2 of a Wheatstone bridge 350 ohm each leg and varies in the 100 microvolts range, thus the need for high gain. The rails, upper and lower voltage are as clean as a battery can be and it only starts oscillating when I connect a load to the FETs, I thought it might have something to do with the inductive coils, so I use a light on one of the power FETs and TP2 rippled, it was flat and varied with the torque like I want and looks great when nothing is attached to the FETs.
TP2 No load TP2 with ~ 2 amp light bulb load and PWM signal No torque on either, however, if enough torque is applied to drive output to 100% TP2 goes smooth again.
Thanks
Kinarfi

 

[BrownOut]

The LM317 is only supposed to provide a couple amps. When you load with ~2 amps, you're already pushing the limit. The motor will be a much bigger load, I'm sure. That's why we keep asking for you to look at the rails.

 

[kinarfi]

The LM317 is only supposed to provide a couple amps. When you load with ~2 amps, you're already pushing the limit. The motor will be a much bigger load, I'm sure. That's why we keep asking for you to look at the rails.

You mean LM324 instead of LM317? and the 2 amp load is through FETs 4 & 5 that is getting a short pulse (TP9 and TP10) (one bulb on each FET drain to + battery)

 

[kinarfi]

Talked to Mike and have redrawn the schematic Now I have to rebuild it. Does anyone have a suggestion of how to couple the output of the PUT oscillator to OP1 and OP2 so I don't load it, This frees up OP4 so I can use it as a buffer for OP3.
Thank You Mike
Thanks all
Kinarfi

 

[MikeMl]

Here is a suggested circuit for the buffering of the PUT Ramp and the means of offsetting it high and low. Note that the offset ramps Hi and Lo get inverted, but I don't think that should matter.