What's digital regulation?
- Thread starter moses
- Start date
Re: What\'s digital regulation?
Never heard of "digital regulation" until Surefire started using the phrase. Must be some term coined by marketing types who know nothing about circuits. I suspect they are using regulation similar to dat2zip's with a micro controller to give some dimming or other functions. Any switcher that I know of is "analog" in nature, of course I may be missing something here. It's probably a misnomer, to say the least.
Never heard of "digital regulation" until Surefire started using the phrase. Must be some term coined by marketing types who know nothing about circuits. I suspect they are using regulation similar to dat2zip's with a micro controller to give some dimming or other functions. Any switcher that I know of is "analog" in nature, of course I may be missing something here. It's probably a misnomer, to say the least.
Entropy
Enlightened
Re: What\'s digital regulation?
Probably it's microcontroller-based.
Not really that much better. A switching regulator is a switching regulator... There are plenty of regulator contollers that will do a MUCH better job than what any current uC can do. (The ATTiny15L from Atmel has some great potential, but it needs a 3-cell config, as under heavy load alkalines would drop below its Vmin of 2.7V)
It's probably just marketing. (BTW, regular switchers COULD be considered "digital" regulation since it primarily deals with a transistor being either full-on or full-off.)
Probably it's microcontroller-based.
Not really that much better. A switching regulator is a switching regulator... There are plenty of regulator contollers that will do a MUCH better job than what any current uC can do. (The ATTiny15L from Atmel has some great potential, but it needs a 3-cell config, as under heavy load alkalines would drop below its Vmin of 2.7V)
It's probably just marketing. (BTW, regular switchers COULD be considered "digital" regulation since it primarily deals with a transistor being either full-on or full-off.)
Jonathan
Enlightened
Re: What\'s digital regulation?
Willie Hunt is a participant in the 'bikecurrent' list, and also an engineer at Surefire. He came up with a regulator for halogen lamps that he calls the 'LVR'. It is really darn simple, and uses a microcontroller driving an FET as the basic switching regulator system.
The microcontroller does the task of regulating the output, using its internal A/D converter and PWM output, and provides higher level functionality such as dimming, soft start, etc. I would bet that the system in the 'Surefire' lights is very similar.
One of the nice features of using a microcontroller with feedback in the digital domain (eg A/D conversion followed by software control laws followed by digital generation of PWM) is that you can have more complex control laws and can adjust these control laws without analog hardware changes.
On the downside, the switching frequency is much lower and the control loops generally much slower. The LVR system works because it isn't a true DC-DC switching regulator, but rather produces a pulsed DC output with the desired RMS voltage. This lets the switching frequency fall below 1 kHz without requiring large inductors or capacitors. However it won't work with LEDs, only with incandescent lamps.
-Jon
Willie Hunt is a participant in the 'bikecurrent' list, and also an engineer at Surefire. He came up with a regulator for halogen lamps that he calls the 'LVR'. It is really darn simple, and uses a microcontroller driving an FET as the basic switching regulator system.
The microcontroller does the task of regulating the output, using its internal A/D converter and PWM output, and provides higher level functionality such as dimming, soft start, etc. I would bet that the system in the 'Surefire' lights is very similar.
One of the nice features of using a microcontroller with feedback in the digital domain (eg A/D conversion followed by software control laws followed by digital generation of PWM) is that you can have more complex control laws and can adjust these control laws without analog hardware changes.
On the downside, the switching frequency is much lower and the control loops generally much slower. The LVR system works because it isn't a true DC-DC switching regulator, but rather produces a pulsed DC output with the desired RMS voltage. This lets the switching frequency fall below 1 kHz without requiring large inductors or capacitors. However it won't work with LEDs, only with incandescent lamps.
-Jon
star882
Enlightened
Re: What\'s digital regulation?
A switching transistor in a SMPS is always completely on or off.
In other words, it is digital!
A switching transistor in a SMPS is always completely on or off.
In other words, it is digital!
Jonathan
Enlightened
Re: What\'s digital regulation?
But in most switching power supplies, the exact time that the switch changes from on to off is _not_ descretized (divided into differentiable discrete levels) as one would expect from a true digital system. I think that one can differentiate between analog on-off modulation and digital on-off modulation.
In analog on-off modulation you do something like compare an analog value to an analog triangle wave, and use that comparison to drive a transistor.
In digital on-off modulation you calculate a value, and at specific points in (the clock cycle of the computer) you drive the transistor on or off as needed.
-Jon
But in most switching power supplies, the exact time that the switch changes from on to off is _not_ descretized (divided into differentiable discrete levels) as one would expect from a true digital system. I think that one can differentiate between analog on-off modulation and digital on-off modulation.
In analog on-off modulation you do something like compare an analog value to an analog triangle wave, and use that comparison to drive a transistor.
In digital on-off modulation you calculate a value, and at specific points in (the clock cycle of the computer) you drive the transistor on or off as needed.
-Jon
![Pampers Sensitive Baby Wipes, Clinically Proven, Fragrance Free, Unscented, Water Based, Hypoallergenic, Multi-Use, Hand Wipes 336 Wipes Total (6X Flip-Top Packs) [Packaging May Vary]](https://m.media-amazon.com/images/I/41KLBHu9CAL._SL500_.jpg)
Re: What\'s digital regulation?
OK, I'm a purist here, semantics notwithstanding. Digital means "everything" is done in the digital domain (except the process of converting the analog to, well, digital). The control loops, ie error generation, loop filtering etc. is done using sampled data techniques. Current monolithic DC-DC converters are not digital in that the "magic" inside is not done in the Z-Domain but in the S-domain. Some may not understand this, but suffice it to say that what's inside the Arc LS, Arc AAA, Surefire KL1, Surefire KL3 are all analog. Now if you throw a uC into the fray, you can stretch the truth a bit (actually a LOT) and claim you now have a digital circuit. BAH!!! The heart of the system is still analog, since the math is described in the continuous time domain. Yes, a switching transistor is digital but all the math that describe power supply behaviour uses S-domain techniques which is, well, analog.
If the circuit in question is analyzed by use of S-domain techniques, it is analog. I bet that the power supply circuits in question are all designed and analyzed using Laplace transform techniques. That means they are analog. A uC is just a cheesy way to claim you have a "digital" product.
Don't get me wrong, a uC is a great way to add flexibility and features into a product. I'm all for having a uC in my light that allows me to tweak things to my heart's content, do flash uploads for cool features, etc. Marketing misnomers just irks me sometimes, that's all. It could well be that Surefire has figured out to do their SMPS *purely* in the digital domain, but I don't think so. It's just not cost effective to do this in the quantities we're talking about. Of course I could be totally 100% wrong here.
Rant mode OFF.
Let the flames begin
OK, I'm a purist here, semantics notwithstanding. Digital means "everything" is done in the digital domain (except the process of converting the analog to, well, digital). The control loops, ie error generation, loop filtering etc. is done using sampled data techniques. Current monolithic DC-DC converters are not digital in that the "magic" inside is not done in the Z-Domain but in the S-domain. Some may not understand this, but suffice it to say that what's inside the Arc LS, Arc AAA, Surefire KL1, Surefire KL3 are all analog. Now if you throw a uC into the fray, you can stretch the truth a bit (actually a LOT) and claim you now have a digital circuit. BAH!!! The heart of the system is still analog, since the math is described in the continuous time domain. Yes, a switching transistor is digital but all the math that describe power supply behaviour uses S-domain techniques which is, well, analog.
If the circuit in question is analyzed by use of S-domain techniques, it is analog. I bet that the power supply circuits in question are all designed and analyzed using Laplace transform techniques. That means they are analog. A uC is just a cheesy way to claim you have a "digital" product.
Don't get me wrong, a uC is a great way to add flexibility and features into a product. I'm all for having a uC in my light that allows me to tweak things to my heart's content, do flash uploads for cool features, etc. Marketing misnomers just irks me sometimes, that's all. It could well be that Surefire has figured out to do their SMPS *purely* in the digital domain, but I don't think so. It's just not cost effective to do this in the quantities we're talking about. Of course I could be totally 100% wrong here.
Rant mode OFF.
Let the flames begin
Jonathan
Enlightened
Re: What\'s digital regulation?
CM,
I happen to agree with you totally.
Just FYI, I can point you to two 'regulator' circuits that I belive qualify as digital.
The LVR that I described uses the A/D converter to measure the battery voltage. It then calculates numerically the inverse square and uses this to adjust the PWM timing of the switching transistor. The output is a duty cycle modulated square wave with the desired RMS voltage.
Ned Konz is looking at the ATTINY15 chip as the core of a boost converter. The input voltage and the output current are measured using A/D conversion, the PWM duty cycle is calculated and sent to a PWM subsystem, and the switching transistor is operated at about 100KHz. The power topology is basic boost converter, but all of the feedback and feed forward is in the digital domain, as is the timing of the switching events.
Oh, for a giggle, look at the output of a 1 bit A/D converter (prior to decimation to multibit words). This is a true digital datastream, in that it has two discreet levels and discreet time transitions. (Of course, producing this digital output is an analog process with various threshold and clocked comparisons, but the digital output is _digital_.) You can take this '1 bit' string of bits at specified time intervals and operate upon it mathematically, eg. decimating to a conventional multibit word digital representation of the original analog signal, storing it on a hard disk, etc. But the neat thing is that this digital signal is _also_ a duty cycle modulated representation of the original analog waveform, and if you simply pass it through an _analog_ low pass filter, you will get your original analog waveform reconstructed
-Jon
CM,
I happen to agree with you totally.
Just FYI, I can point you to two 'regulator' circuits that I belive qualify as digital.
The LVR that I described uses the A/D converter to measure the battery voltage. It then calculates numerically the inverse square and uses this to adjust the PWM timing of the switching transistor. The output is a duty cycle modulated square wave with the desired RMS voltage.
Ned Konz is looking at the ATTINY15 chip as the core of a boost converter. The input voltage and the output current are measured using A/D conversion, the PWM duty cycle is calculated and sent to a PWM subsystem, and the switching transistor is operated at about 100KHz. The power topology is basic boost converter, but all of the feedback and feed forward is in the digital domain, as is the timing of the switching events.
Oh, for a giggle, look at the output of a 1 bit A/D converter (prior to decimation to multibit words). This is a true digital datastream, in that it has two discreet levels and discreet time transitions. (Of course, producing this digital output is an analog process with various threshold and clocked comparisons, but the digital output is _digital_.) You can take this '1 bit' string of bits at specified time intervals and operate upon it mathematically, eg. decimating to a conventional multibit word digital representation of the original analog signal, storing it on a hard disk, etc. But the neat thing is that this digital signal is _also_ a duty cycle modulated representation of the original analog waveform, and if you simply pass it through an _analog_ low pass filter, you will get your original analog waveform reconstructed
-Jon
Re: What\'s digital regulation?
One might need to clarify that Willie Hunt was working on his LVRs way befor he joined SF - one might even suspect that SF called him to work on exactly those digital regulation parts of the new designs because of his known work on these LVRs.
Wiillie Hunt LVR site
Willie Hunt main site
Klaus
One might need to clarify that Willie Hunt was working on his LVRs way befor he joined SF - one might even suspect that SF called him to work on exactly those digital regulation parts of the new designs because of his known work on these LVRs.
Wiillie Hunt LVR site
Willie Hunt main site
Klaus
Re: What\'s digital regulation?
Jon, I would qualify those as "digital" if processing is done in Z-domain. Sounds like they are which is pretty cool since it is a departure from convention. I've played with 1bit DAC's at work and they are pretty neat. You get an analog waveform right out of a chip that puts out 1's and zeroes. Of course you need the reconstruction filter which is nothing more than a cheap RC network.
Jon, I would qualify those as "digital" if processing is done in Z-domain. Sounds like they are which is pretty cool since it is a departure from convention. I've played with 1bit DAC's at work and they are pretty neat. You get an analog waveform right out of a chip that puts out 1's and zeroes. Of course you need the reconstruction filter which is nothing more than a cheap RC network.
Jonathan
Enlightened
Re: What\'s digital regulation?
Showing my ignorance here, but I thought that the Z domain was 'simply' the discrete time S domain. I'd certainly appreciate it if you would expand on the difference.
-Jon
Showing my ignorance here, but I thought that the Z domain was 'simply' the discrete time S domain. I'd certainly appreciate it if you would expand on the difference.
-Jon
bikeNomad
Newly Enlightened
Re: What\'s digital regulation?
The gate of a typical power FET has a high capacitance, which has to be charged or discharged to have it change states. Even the little SO-8 N-ch FETs (like the Si9410DY) that I've been using can have a gate charge of 20 nanocoulombs (nC) or so (Vgs=5V, Vdd=15V).
The faster you change the voltage on the FET gates, the more current it takes. Which means that you have to have bigger transistors driving the gates, or put up with slower transitions.
The downside of these slower transitions is that while the power transistor is in its "analog" mode (i.e. not fully ON and not fully OFF), it's dissipating more heat (wasting more energy).
So imagine charging or discharging the 20 nC gate charge with a microcontroller PWM output that can only source/sink 20mA. This would take 1 usec.
At a PWM rate of 100KHz (10usec), this means that fully 20% of the time the transistor is somewhere between fully ON and fully OFF!
Which, of course, results in more energy being wasted as heat.
And don't get me started on Rds(ON), the on resistance of the FETs...
<font size="2" face="Verdana, Arial">I wish that were the case. The truth is that the time (and current) it takes to get from ON to OFF or vice versa can cause some significant losses.
The gate of a typical power FET has a high capacitance, which has to be charged or discharged to have it change states. Even the little SO-8 N-ch FETs (like the Si9410DY) that I've been using can have a gate charge of 20 nanocoulombs (nC) or so (Vgs=5V, Vdd=15V).
The faster you change the voltage on the FET gates, the more current it takes. Which means that you have to have bigger transistors driving the gates, or put up with slower transitions.
The downside of these slower transitions is that while the power transistor is in its "analog" mode (i.e. not fully ON and not fully OFF), it's dissipating more heat (wasting more energy).
So imagine charging or discharging the 20 nC gate charge with a microcontroller PWM output that can only source/sink 20mA. This would take 1 usec.
At a PWM rate of 100KHz (10usec), this means that fully 20% of the time the transistor is somewhere between fully ON and fully OFF!
Which, of course, results in more energy being wasted as heat.
And don't get me started on Rds(ON), the on resistance of the FETs...
Re: What\'s digital regulation?
http://sepwww.stanford.edu/sep/prof/fgdp/c2/paper_html/node9.html
<font size="2" face="Verdana, Arial">You can go from one domain to the other using Z-transforms. One common transformation is the bilinear transform which is a mapping from the s-plane to the Z-plane. The math gets kinda involved but I found a link that describes the transform if you're interested. The mapping takes one from a unit circle to left/right half plane descriptions. This is for propellerheads only
http://sepwww.stanford.edu/sep/prof/fgdp/c2/paper_html/node9.html
INRETECH
Flashlight Enthusiast
Re: What\'s digital regulation?
In switching regulators, there are several different "modes" of operation - they basically try to adjust the PWM of the control signal to keep the output voltage at the desired level
Current, Voltage, Ripple and Cuk'
Some are sync and some are non-sync
With the inductor, you have to make sure its large enough to keep the regulator from entering discontious mode
With the cap, you have to make sure it can handle the ripple current
I once debugged a problem with a switching power supply that the caps kept blowing up, and it was NOT:
) Overvoltage
) Over current
) Reverse polarity
) Too high ripple current
Lets see if anyone can guess why ?
In switching regulators, there are several different "modes" of operation - they basically try to adjust the PWM of the control signal to keep the output voltage at the desired level
Current, Voltage, Ripple and Cuk'
Some are sync and some are non-sync
With the inductor, you have to make sure its large enough to keep the regulator from entering discontious mode
With the cap, you have to make sure it can handle the ripple current
I once debugged a problem with a switching power supply that the caps kept blowing up, and it was NOT:
) Overvoltage
) Over current
) Reverse polarity
) Too high ripple current
Lets see if anyone can guess why ?
Kercheval
Newly Enlightened
Re: What\'s digital regulation?
Too frequent large voltage swing charge/discharge cycle rates resulting in too large a heat dissipation requirement?
Too frequent large voltage swing charge/discharge cycle rates resulting in too large a heat dissipation requirement?
INRETECH
Flashlight Enthusiast
Re: What\'s digital regulation?
They were Electrolytic and they were operating well within their ratings
Nothing to do with heat, or vibration either
We got back several of these power supplies and just didn't think much of it until one of them caught fire on the district manager's desk
I had a terrible time with security wanting me to sign a property pass as I was trying to take it outside the building and dump it in a trash container
They were Electrolytic and they were operating well within their ratings
Nothing to do with heat, or vibration either
We got back several of these power supplies and just didn't think much of it until one of them caught fire on the district manager's desk
I had a terrible time with security wanting me to sign a property pass as I was trying to take it outside the building and dump it in a trash container
star882
Enlightened
Re: What\'s digital regulation?
The caps are bad.
The caps are bad.
INRETECH
Flashlight Enthusiast
Re: What\'s digital regulation?
No, the caps were perfectly alright
I can take a normal brand-new Electrolytic cap, run it at its operating parameters and make it explode
No, the caps were perfectly alright
I can take a normal brand-new Electrolytic cap, run it at its operating parameters and make it explode
