Author Topic: #225 - Added Thoughts  (Read 8966 times)

0 Members and 1 Guest are viewing this topic.

Offline 45Overload

  • Contributor
  • Posts: 35
  • Country: us
#225 - Added Thoughts
« on: December 06, 2011, 05:13:55 pm »
Thanks! Interesting info on PWM circuits and power supply control.  No question about it, PWM is a very important subject and it works very well in lots of applications.  My thinking on power supply tutorials is that they should be attractive to beginning builders.  Using basic building blocks and discovering what works and what doesn't is essential.  PWM control of your bench supply goes far beyond this level and is overkill to say the least. 

First, a good 10-turn pot might cost $200.00 for a good quality one.  You can get used ones or surplus on ebay for 10-20 bucks and hope they don't have dead spots.  But frankly they are an excellent and simple way to get the adjustment precision that you want in a simple and direct manner.

Let's mention rotary encoders briefly.  The 10 cent encoder shown in the video is used on nearly all consumer grade electronics to adjust volume, step through menus, or any adjustment requiring a digital input to a programmed micro processor.  This part uses mechanical wipers to create a quadrature signal for the micro.  The mechanical wipers last a few months if you are lucky before they start to wear out and give switch bounce.  They are junk.  But, since they are cheap you can just keep replacing them if that's ok with you.  The alternative is - you guessed it - an expensive optical encoder costing about the same as a good quality 10-turn control.  $100 to $200 will get you a reasonable quadrature optical encoder that will last forever.  Keep in mind that you have to feed the quadrature signal into a specially programmed PIC or similar that interprets the data, sets the rate of increase or decrease and can also memorize the last used position.  Then you have other choices besides the use of PWM.  The D/A output can come from a look-up table, or a formula or whatever.  But the point is why make everything into a software programming problem? 

Do I want to construct a pc board for a microprocessor and its supporting elements just to have a digitally controlled output on my power supply.  No thanks.  I'm not saying PWM isn't effective and a great tool.  That's what I would use if the power supply was going to be produced in quantity - that way the micro can do lots of things besides control the voltage.  But I thought the purpose of the videos was to encourage the first time builder to construct a useful part of his repair bench. 

PWM is used extensively for high power control circuits such as motor drives, especially industrial 3-phase motor controllers.  PWM control, along with IGBT or FET's  is "state of the art" in these high power control applications.  For a simple bench supply, nothing beats a 10-turn pot, some voltage dividers and a couple of op-amps in a feedback loop.  SOMEBODY has to do the engineering, somewhere.  Otherwise forget the soldering and just start writing code.  My 2 cents worth.

45O
 

Offline deephaven

  • Frequent Contributor
  • **
  • Posts: 783
  • Country: gb
  • Civilization is just one big bootstrap
    • Deephaven Ltd
Re: #225 - Added Thoughts
« Reply #1 on: December 06, 2011, 05:26:00 pm »
$25 will get you a good quality Bournes optical incremental encoder, so it's not that expensive. Dave was sufficiently detailed in his videos that a newcomer should be able to work out where a simple pot goes if he doesn't want to go to the complexity of putting a micro in there.

Giving people options provides a lot more depth to the blog and encourages people to use their imagination rather than the "Thall shall do it this way" approach.
 

Online IanB

  • Super Contributor
  • ***
  • Posts: 9561
  • Country: us
Re: #225 - Added Thoughts
« Reply #2 on: December 06, 2011, 05:53:59 pm »
I did notice when Dave was explaining about the lack of precision in a single turn pot, he tried to turn the spindle with his fingers with the pot floating loose. This is like taking great care over electronic design but paying no attention to mechanical design. It would of course be possible to get better precision if the pot were fixed in place and had a large knob affixed. To some extent the turn precision is proportional to the diameter of the knob. A knob with a large diameter works way better than a bare spindle for fine control. (There are limits, obviously, but one might as well take best advantage of whatever precision is available in the system.)
I'm not an EE--what am I doing here?
 

Offline firewalker

  • Super Contributor
  • ***
  • Posts: 2328
  • Country: gr
Re: #225 - Added Thoughts
« Reply #3 on: December 06, 2011, 06:24:35 pm »
Don't put aside the educational aspect of those videos. Yes, someone could stop at the usage of a pot. Is there an alternative? Yes. PWM and rotary encoders with an mcu is one.

An "old" engineer can make the decisions by him self (he wouldn't even need to watch such a tutorial video). He knows what he needs.

It is very good for a student etc to extend his options.

Alexander.
Become a realist, stay a dreamer.

 

Offline FreeThinker

  • Frequent Contributor
  • **
  • Posts: 791
  • Country: england
  • Truth through Thought
Re: #225 - Added Thoughts
« Reply #4 on: December 06, 2011, 06:36:29 pm »
I did notice when Dave was explaining about the lack of precision in a single turn pot, he tried to turn the spindle with his fingers with the pot floating loose. This is like taking great care over electronic design but paying no attention to mechanical design. It would of course be possible to get better precision if the pot were fixed in place and had a large knob affixed. To some extent the turn precision is proportional to the diameter of the knob. A knob with a large diameter works way better than a bare spindle for fine control. (There are limits, obviously, but one might as well take best advantage of whatever precision is available in the system.)

The main problem with a single turn pot is it's non linearity ie for a 300 degree turn pot @ 300 ohms you should expect 1ohm/degree but in practice you will get nothing like this. Even adding a large knob will not help. While a ten turn pot is still not linear the fact that it is ten turns greatly helps in even out the lumps and bumps. If you want a really bad pot for tracking try a single turn Log pot. They really should be questioned under the trades descriptions act as they are anything but log :D. An old trick was to use a single turn pot feeding a log law opamp to get better linearity (or would that be logarity??) but it still wasn't great.
Machines were mice and Men were lions once upon a time, but now that it's the opposite it's twice upon a time.
MOONDOG
 

Online IanB

  • Super Contributor
  • ***
  • Posts: 9561
  • Country: us
Re: #225 - Added Thoughts
« Reply #5 on: December 06, 2011, 06:45:34 pm »
The main problem with a single turn pot is it's non linearity ie for a 300 degree turn pot @ 300 ohms you should expect 1ohm/degree but in practice you will get nothing like this.

IMHO linearity is not critical if you have feedback. For instance, if you are trying to tweak the pot to get exactly 5.00 V on the display what you really need is fine adjustment control. I once reclaimed a giant control knob from some old piece of equipment and it was fantastic for fine adjustment of a plain old regular pot.
I'm not an EE--what am I doing here?
 

Offline FreeThinker

  • Frequent Contributor
  • **
  • Posts: 791
  • Country: england
  • Truth through Thought
Re: #225 - Added Thoughts
« Reply #6 on: December 06, 2011, 07:12:50 pm »
Linearity is not critical as you say but goes a long way to help with sensitivity. I have used mechanically geared pots with great effect but they cannot improve on the basic linearity of the pot only stretch its imperfections over a wider scale. If you are using a pot to set your output volts and you cannot easily select say 5volts no amount of feedback will get you your required output, it will only clamp your  output. A larger knob or a mechanically geared knob help you select your required output with more accuracy but come nowhere near a decent 10 turn pot. Add a vernier scale and you have a much better and simpler alternative to the mechanically geared pot at a much more reasonable cost.
« Last Edit: December 06, 2011, 07:15:39 pm by FreeThinker »
Machines were mice and Men were lions once upon a time, but now that it's the opposite it's twice upon a time.
MOONDOG
 

alm

  • Guest
Re: #225 - Added Thoughts
« Reply #7 on: December 06, 2011, 07:23:26 pm »
Limited settability is an issue, I've had issues with supplies refusing to be set to a certain value (eg. 5.00 V), or quickly jumping from from 0.7 to 1.5 V (linearity) (this was 30V full scale). No amount of mechanical improvements are going to help with this. Of course many of the multiturn pots are wirewound, which has its on resolution limit due to their construction (each turn of the wire is one discrete step).
 

Offline FreeThinker

  • Frequent Contributor
  • **
  • Posts: 791
  • Country: england
  • Truth through Thought
Re: #225 - Added Thoughts
« Reply #8 on: December 06, 2011, 07:38:59 pm »
Limited settability is an issue, I've had issues with supplies refusing to be set to a certain value (eg. 5.00 V), or quickly jumping from from 0.7 to 1.5 V (linearity) (this was 30V full scale). No amount of mechanical improvements are going to help with this. Of course many of the multiturn pots are wirewound, which has its on resolution limit due to their construction (each turn of the wire is one discrete step).
This is the beauty of digital switches as you can design in the step size with your dac ( Or with pwm ) but you still can't get infinite resolution  :'(. But 10mV/ step is good enough for all but the most demanding design and 1.2mV / step is just about perfect  8). Very Nice design so far but I suspect it will be improved even further before committing to pcb. Vague promise of up to 10v in another thread but we will have to wait and see. Certainly brought a sparkle to my Christmas.
Machines were mice and Men were lions once upon a time, but now that it's the opposite it's twice upon a time.
MOONDOG
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29706
  • Country: au
    • EEVblog
Re: #225 - Added Thoughts
« Reply #9 on: December 06, 2011, 10:07:45 pm »
Do I want to construct a pc board for a microprocessor and its supporting elements just to have a digitally controlled output on my power supply.  No thanks.  I'm not saying PWM isn't effective and a great tool.  That's what I would use if the power supply was going to be produced in quantity - that way the micro can do lots of things besides control the voltage.  But I thought the purpose of the videos was to encourage the first time builder to construct a useful part of his repair bench. 

That's why I went to the trouble to explain that there are various ways to control your power supply, and what a great thing 10 turn pots are. Only in the 3rd video did I introduce PWM as an option if people want to do that.
The circuit has the advantage of being able to seamlessly change between POT control or PWM/DAC control with no changes, entirely the constructors choice.

Dave.
 

Offline FreeThinker

  • Frequent Contributor
  • **
  • Posts: 791
  • Country: england
  • Truth through Thought
Re: #225 - Added Thoughts
« Reply #10 on: December 06, 2011, 11:30:10 pm »
From a cost/performance point of view (especially in low volume/ one off) the ten turn pot is a clear winner. The Pwm+ controller has the advantage of flexibility and the encoder /dac the convenience of ease of implementation and robustness. Each has its place and other design considerations will dictate which one is employed. I have a power supply that I was going to mod to take ten turn pots. The first problem was limited space, ten turn pots tend to be quite deep and so foul the front panel/pcb. Testing them off board showed that while I indeed got more turn on the pot before the digits incremented on the meter I still had no more control over the output due to the resolution of the meter (not surprising) but also even with an external meter could not get better than the 10mv resolution of the original meter. So in the end I left it as original, the single turn pots are not very sensitive  but with care can be set to the supplies minimum increment. Ten turn pots are easier if you have the space but in this case were really overkill
Machines were mice and Men were lions once upon a time, but now that it's the opposite it's twice upon a time.
MOONDOG
 

alm

  • Guest
Re: #225 - Added Thoughts
« Reply #11 on: December 07, 2011, 12:02:35 am »
Useful resolution also depends on the stability of the power supply. Not much point in <10 mV settability if the output drifts by a few mV per minute.
 

Offline Zad

  • Super Contributor
  • ***
  • Posts: 1013
  • Country: gb
    • Digital Wizardry, Analogue Alchemy, Software Sorcery
Re: #225 - Added Thoughts
« Reply #12 on: December 07, 2011, 03:30:09 am »
It would be nice to know what Dave's intentions are for this power supply. I guess there will be more videos in the series - a great subject for a set of CAD videos on schematic capture, library component design, and layout. After that, maybe selling bare PCBs, kits or populated PCBs? I reckon Circuit Cellar / Elektor would be interested in publishing it, which would send in lots of new YouTube viewers.

With the chat about SMUs on this weeks Amp Hour, how about incorporating current and voltage measurement? Even cheap micros have a couple of onboard ADCs with sufficient capability for a product like this.


Offline BravoV

  • Super Contributor
  • ***
  • Posts: 6113
  • Country: 00
Re: #225 - Added Thoughts
« Reply #13 on: December 07, 2011, 05:14:17 am »
IMHO linearity is not critical if you have feedback. For instance, if you are trying to tweak the pot to get exactly 5.00 V on the display what you really need is fine adjustment control. I once reclaimed a giant control knob from some old piece of equipment and it was fantastic for fine adjustment of a plain old regular pot.
Limited settability is an issue, I've had issues with supplies refusing to be set to a certain value (eg. 5.00 V), or quickly jumping from from 0.7 to 1.5 V (linearity) (this was 30V full scale). No amount of mechanical improvements are going to help with this. Of course many of the multiturn pots are wirewound, which has its on resolution limit due to their construction (each turn of the wire is one discrete step).
Agree with Alm comment, on using single turn pot, no matter how big the knob is used, there is a limitation and I've been struggling with it. Thats the reason sometimes for cheap solution is using two single turn pots for coarse and fine tuning.

Currently I'm working on a project that needs a physical pot and as Alm said, even a high quality wirewound 10 turns pot has the limitation on it's fine resolution, that makes me have to use two 10 turns pots for coarse and fine.  :-\

Just look at the examples below and watch the size, the common cheap single turn pot like at the right simply can not beat the big brothers 10 turns pots like these, with the 0.1% linearity value, that cheap pot is not even in the same league.

« Last Edit: December 07, 2011, 05:36:27 am by BravoV »
 

Online IanB

  • Super Contributor
  • ***
  • Posts: 9561
  • Country: us
Re: #225 - Added Thoughts
« Reply #14 on: December 07, 2011, 05:22:29 am »
Agree with Alm comment, on using single turn pot, no matter how big the knob is used, there is a limitation and I've been struggling with it.

I know there's a limit, but I'm just saying that turning the bare spindle with your fingers is not going to get the best out of whatever limited precision is available. It's when you need to go beyond the inherent limits of a simple cheap pot that the more expensive devices become necessary.
I'm not an EE--what am I doing here?
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29706
  • Country: au
    • EEVblog
Re: #225 - Added Thoughts
« Reply #15 on: December 07, 2011, 05:28:41 am »
It would be nice to know what Dave's intentions are for this power supply. I guess there will be more videos in the series - a great subject for a set of CAD videos on schematic capture, library component design, and layout. After that, maybe selling bare PCBs, kits or populated PCBs? I reckon Circuit Cellar / Elektor would be interested in publishing it, which would send in lots of new YouTube viewers.

Yes, the intention is for it to become a kit.
As for the final feature set, I have a few tricks left up my sleeve I haven't shown or discussed yet, that will make it, IMO, a pretty unique supply.

Dave.
 

Offline BravoV

  • Super Contributor
  • ***
  • Posts: 6113
  • Country: 00
Re: #225 - Added Thoughts
« Reply #16 on: December 07, 2011, 05:48:57 am »
Yes, the intention is for it to become a kit.
As for the final feature set, I have a few tricks left up my sleeve I haven't shown or discussed yet, that will make it, IMO, a pretty unique supply.

Dave.

Dave, sorry for oot, is there any plan in the process of procuring the kits for this power supply, also includes the procurement for the ucurrent kits in the process ?  ;)

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29706
  • Country: au
    • EEVblog
Re: #225 - Added Thoughts
« Reply #17 on: December 07, 2011, 06:02:06 am »
Dave, sorry for oot, is there any plan in the process of procuring the kits for this power supply, also includes the procurement for the ucurrent kits in the process ?  ;)

You've missed the updates, the uCurrent kits are at the assemblers now.

Dave.
 

Offline BravoV

  • Super Contributor
  • ***
  • Posts: 6113
  • Country: 00
Re: #225 - Added Thoughts
« Reply #18 on: December 07, 2011, 06:11:56 am »
Thanks, guess I missed that announcement, great, cant wait to pull the trigger on buying that gadget.

Offline sonicj

  • Frequent Contributor
  • **
  • Posts: 752
  • Country: us
  • updata successed!
Re: #225 - Added Thoughts
« Reply #19 on: December 07, 2011, 08:46:39 am »
i thoroughly enjoyed the series! seeing the thought process play out in real time helped me to connect a lot of concepts that i vaguely understood prior.

i personally liked how dave included detailed explanations for both methods of input control.  there are advantages & disadvantages to both and i think he did a great job articulating that. i came across a good example of this in a real world application just yesterday...  my dj mixer uses resistive fader pots for both input channels, but a optical fader and ┬ÁC for the often abused cross-fader.  from the outside of the case, all 3 faders look identical and even perform similar tasks. under real world conditions, their mode of operation is very, very different.
-sj
 

Offline FreeThinker

  • Frequent Contributor
  • **
  • Posts: 791
  • Country: england
  • Truth through Thought
Re: #225 - Added Thoughts
« Reply #20 on: December 07, 2011, 12:03:37 pm »
i thoroughly enjoyed the series! seeing the thought process play out in real time helped me to connect a lot of concepts that i vaguely understood prior.

-sj
The most illuminating part for me was the thought processes behind the design for example the led on the output. He shows you a perfectly reasonable design  that unfortunately does not work as expected explains why it fails and shows how to make it perform as required. If this was intentional or not I don't know but it illustrates beautiful how designs change as they develop and why breadboarding is so good for the job, but at the same time can produce it's own set of problems (voltage drop, stray capacitance etc) great example of how a design is tweaked as it goes along and a pointer as to why he has so many unfinished projects, he will always find that little extra tweek in the search for perfection :D   
Machines were mice and Men were lions once upon a time, but now that it's the opposite it's twice upon a time.
MOONDOG
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: #225 - Added Thoughts
« Reply #21 on: December 07, 2011, 01:12:41 pm »
Nice photos.  You can find high quality pots in salvage medical grade or avionics grade equipment.  The pots alone are worth more than the selling price of these eBay 'for parts or not working' types.


Just look at the examples below and watch the size, the common cheap single turn pot like at the right simply can not beat the big brothers 10 turns pots like these, with the 0.1% linearity value, that cheap pot is not even in the same league.

Best Wishes,

 Saturation
 

Online don.r

  • Frequent Contributor
  • **
  • Posts: 673
  • Country: ca
Re: #225 - Added Thoughts
« Reply #22 on: December 07, 2011, 03:06:30 pm »
I like the flexibility of a ucontroller in this application. No need to use a rotary encoder if you don't want to as a pair of momentary switches or a momentary rocker will do as well. Arduino's are $15 on ebay so even novices can learn to program and be up and running in a few minutes. Everyone will need to learn how to design with and program an MCU at some point so might as well do it on a simple project like this first.
Shippers: for the love of Pete and all that is holy, STOP USING UPS INTERNATIONAL!
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: #225 - Added Thoughts
« Reply #23 on: December 07, 2011, 04:20:08 pm »
One item to be mindful off in making a DC power supply with digital controller, such as PWM, is the noise injected to the output. 

"Cleanliness" can be done, but I've not seen a manufacturer make a lab PSU with a digital controller cheaper than and meet the low noise capacity of, an equivalent all analog linear supply.   Chances are one's home-brew PWM controlled supply is going to be noisier than if it were designed in purely analog fashion.

For example only because its online, here are specs for Instek.  Just estimating my recollection, you can get a linear supply with 10x less noise than a digitally controlled linear supply for typically 5-10x less cost:

Programmable linear supplies:

http://www.gwinstek.com/en/product/productdetail.aspx?pid=38&mid=52&id=117

vs old fashion analog supplies:
http://www.gwinstek.com/en/product/productdetail.aspx?pid=38&mid=53&id=158

« Last Edit: December 07, 2011, 04:23:35 pm by saturation »
Best Wishes,

 Saturation
 

Offline BravoV

  • Super Contributor
  • ***
  • Posts: 6113
  • Country: 00
Re: #225 - Added Thoughts
« Reply #24 on: December 07, 2011, 05:14:15 pm »
Nice photos.  You can find high quality pots in salvage medical grade or avionics grade equipment.  The pots alone are worth more than the selling price of these eBay 'for parts or not working' types.

Yeah, I've seen their internal, quite amazing piece of work, actually I have another one coming on it's way, and the seller told me it was pulled from a communication panel from a US military spy plane from cold war era, not sure if he is bs-ing me or not, but physically its significantly bigger and fatter than those three which are already considered quite big, but this one has a 0.01% linearity which is ten times better, just wish me luck and hope it arrived safely.  :-\


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf