classic 150 questions and suggestions

[boB]

Yes, with = wish   

Not sure what you mean by MSD in this context ?

0.05 volt resolution would be the goal.  From a longer time average, if the 0.1 volt resolution were real accurate, we could probably derive an in-between average of 0.05V resolution maybe.

Think of a battery voltage for example toggling between, say, 48.3 V and 48.4 V at 50%...  That could be interpreted as 48.35 volts

[Barry Fields]

Like I say I am no engineer. I only have a cursory knowledge of the MCU used in the AIMs inverter.

If the following is of assistance to you, great.

MOTOROLA.COM/SEMICONDUCTORS
M68HC08
Microcontrollers
MC68HC908MR32/D
Rev. 6.0
MC68HC908MR32

3.3.2 Voltage Conversion
When the input voltage to the ADC equals VREFH , the ADC converts the signal to
$3FF (full scale). If the input voltage equals VREFL, the ADC converts it to $000.
Input voltages between VREFH and V REFL are straight-line linear conversions. All
other input voltages will result in $3FF if greater than V REFH and $000 if less than
VREFL.

My sense is you have no reason to accurately measure Bank voltages less than 3 or 4 volts.

I have no idea if your MCU has similar capabilities.

Just throwing it out there.
Barry
 

[boB]

I used the  M68HC08 line of controllers 30+ years ago.  They were fine for the time.

Then became Freescale and now, Microchip ?  I can't keep up with names and who bought who.

Another way to read battery voltages more accurately with a lower bit A/D converter would be to bias the input voltage down so that the low end you are measuring, say a 48V battery only reading down to maybe 35 volts, becomes zero for the ADC.

The Classic works from around 9 volts up to 100 V DC or so, so that can't be done enough to help its resolution.

Interesting engineering issues and I do like discussing these.

boB

[Barry Fields]

boB,
Could you share with me which MCU the Classic 150 uses?  I might then better educate my self and ask less dumb questions.

I have to stop dreaming about the 150 and get back to Gisele Bündchen.

Barry

[boB]

The Classic uses an  NXP   LPC2366  ARM chip.  Old technology by today's standards.

Not even a Cortex-M   It was/is still a good processor but if I were changing it, I would probably go to an STM32F4xx  Cortex M4 part.

boB

[Barry Fields]

boB,
Thanks for the info. When I say something wrong, please correct me. These suggestions are made trying to avoid a total redesign of the 150.

As I read it there are 6 ADC inputs on pins 23, 24, 25, 26, 30 and 31.  I do not know know which are used or might still be available.

My presumption is you are using ONE precision voltage divider on one ADC input to automatically determine BANK voltage and to control BANK charging parameters.  This one divider is aimed at a 96volt Bank.

It would dramatically improve resolution if each BANK voltage had it's own divider.

If one more ADC input were available, It could be used to control BANK charging parameters. A user installable (plug in) BANK voltage specific voltage divider could solve most resolution issues. Unfortunately this approach is NOT IDIOT PROOF. The wrong divider could cause MCU POOF. I'm not sure how many 3.3v zeners would be required to preclude this.

IDIOT PROOFING would require an addition of a HV addressable analog switch. That would depend on remaining MCU resources and room inside "the box" for a jack board.

It might be that only two additional dividers are necessary. The existing divider used for 72v and 96v BANKS, another for 32v and 48v BANKS , and the third for 12v and 24v BANKS.

[Barry Fields]

Me thinks I should retreat to the sidelines for now. Kinda feel I have thrown enough tennis balls to your side of the net to ponder.

I want to thank boB, Ryan, Vic and Larry for their contributions to make this post exceed 1000 views. All this Newby did was ask a few questions.

That 1000 views should be taken as an indicator of the interest in the future of the Classic 150.

I remain at your service,
Barry

[Barry Fields]

Some where in the myriad of posts I have perused, I remember a request for averaging the Wizbang current measurement.

Would the addition of a 30-60 second Low leakage R/C network to the Wizbang output be a solution?

Just asking.
Barry

[boB]

There is already an R-C filter on the WB Jr. but not 30 to 60 seconds.

 Enough filtering to stop fast current spikes though.

30+ seconds would be way too much filtering.

A digital filter could of course keep the fast captures while doing what you are thinking.  You can also request the WB Jr. samples continuously, 10 times per second through modbus. Then filter by external computer such as a Raspberi Pi.

Then, if wanted, that filtered information could be fed back to the Classic if you wanted to control it a bit differently based on that data.   That is what is pretty cool about communications...

A fun project !

boB

[Barry Fields]

I must have misunderstood. I thought the complaint was a premature dropout of ABSORB at end amps.
I suppose if the end amps reading is validated by an ABSORB setpoint voltage, then it should not be a problem.
Sorry for the misunderstanding.
Barry

[boB]

I must have misunderstood. I thought the complaint was a premature dropout of ABSORB at end amps.
I suppose if the end amps reading is validated by an ABSORB setpoint voltage, then it should not be a problem.
Sorry for the misunderstanding.
Barry

Oh !   THAT problem !

I understand that one.  I will look to see if the newer firmware changes fixed that or not.

The problem was that when the Classic is sitting there in Absorb below the ending-amps setting, regulation or loads will change such that the current rises just above the ending-amp setting momentarily and resets the ending-amps timer which is 60 seconds or 90 seconds.

That is a software issue but might be fixed also by a LARGE R-C network.