Rethinking our approach to the DDL?

[Wolfman29]

While, I am not going to get involved in this argument (well, I guess I will...), I think the idea of asking for help is perfectly legitimate. RHD helps a lot here, and if he needs a bit of held in return, why not give it to him?

Anyway, I really hope this circuit works out... if I have time tomorrow (spending time with my girl tomorrow, so may not), I will try to test out this current sink on LTSpice.

 

[HIMNL9]

I'm still wondering why you are insisting with voltage regulators for current drivers

There are CURRENT-based regulators, that can be used for linear drivers, with very low total dropout (meaning, the dropout that all the driver cause), that can be efficently used already .....

Like the LT3080 for less than 1A, LT3085 for less than 500mA and the LT3083 for over 1A, as example ..... and they can use a very low RSET value (in fact, lower than in the datasheets), and can be paralleled in case of needs, and can be made regulable using a very low power trimmer .....

 

[Wolfman29]

Hmm. You may be right. I actually have a sample of the LT3083 (stumbled upon it a while ago), just never managed to get it to work. I may tinker with it as well if I find the time.... I will report findings when I get them.

 

[HIMNL9]

Well, it works decently good, for me, but you need to remember that they are current sources, not sinks ..... (anyway, it can work as "pseudo-sink", if needed), and the dropout, in the way we use it, it's always the bigger one (it's indicated in the datasheet, that the dropout is 310mV ONLY if the Vcontrol is independent (and higher) than the Vin, otherwise the total dropout is still the one of the Vin ..... apart this, you can use it with a 0.15 ohm RSET and a 10K trimmer (this need some extra capacitors at in and out, for the stability), and regulate the current til 3A ..... in these conditions, the TOTAL dropout of the assembly is like, as example, 1.5V for 1.8 to 2 A, not too bad, compared with other systems

BTW, also remember that they don't work correctly in open-loop conditions, being "pass-type" regulators for work correctly they need to have a minimum of 1mA current absorption from the load (anyway, a common 400 / 500 ohm resistor in parallel to the output is more than enough, usually)

 

[lasersbee]

Not to dissuade anyone from reading, but there's such a thing as unnecessary reading. If there's a common part number that someone can reply with in 8 characters, so that I can order it now, and then figure it out in three weeks when it actually arrives from China and I have more time, why would it be so offensive to you that I ask for this courtesy instead of spending 30 minutes on digikey reading datasheets now?

Well Jerry, you are trying to offend someone. Saying that you're not, at the end of your thread, doesn't change the contents of what you've actually said.

If we're going to be honest here, I can't really remember the last time that I've read a thread where I left feeling as though YOU had contributed something positive to the dialogue or debate.

I cannot say that I get the same impression from you. I rarely observe you doing anything positive, or encouraging free open discussion. I all the more often observe you trying to shut down ideas, and maintain the status quo. What is it about this forum that attracts you Jerry? You've certainly been here a lot longer than I. I don't get it.

DigiKey Data sheets are free and easy to access.... why should
I or any other member spend their time reading them for you..
when it seems to be too much of a bother for you to do it yourself...:thinking:

It is obvious that I DID offend you.... and for that I apologize...

So now who is trying to offend whom....
I know what I contribute to the forum in the form of expertise
and knowledge...
I'm sorry if those contributions are not what you want to hear
or read....
And no.... You did not offend me with that long winded reply...

I'm sure you will find the Holy Grail of LDO Linear Voltage
Regulators that you are looking for...:beer:


Jerry

 

[random person]

DigiKey Data sheets are free and easy to access.... why should
I or any other member spend their time reading them for you..
when it seems to be too much of a bother for you to do it yourself...:thinking:
Jerry

........... =P Gosh... cmon people.... conflict is so....lame ...can't we move past that..? If your not going to help just don't bother replying. Its what I do if i'm annoyed. To Rhd's defense, there a 100,000+ IC models out there. I'll attempt to help you guys out in a few hours. Gotta get back to work. Can also show you a bjt+mosfet current limiting schematic if your interested. BBL

 

[Benm]

I'm still wondering why you are insisting with voltage regulators for current drivers

There are CURRENT-based regulators, that can be used for linear drivers, with very low total dropout (meaning, the dropout that all the driver cause), that can be efficently used already .....

Like the LT3080 for less than 1A, LT3085 for less than 500mA and the LT3083 for over 1A, as example ..... and they can use a very low RSET value (in fact, lower than in the datasheets), and can be paralleled in case of needs, and can be made regulable using a very low power trimmer .....

Well, not to be too picky, but a LM3083 is still a voltage regulator by nature. To make it into a current regulator you pull the exact same trick you do with a LM317: rig a sense resistor between output pin and reference pin, and connect the LD after the sense resistor.

The only difference is that it will work with much much lower sense values than the defined 1.25 volts for the LM317, and that you can control the sense voltage with a potmeter (instead of adjusting a potmeter in the middle of the full current path).

Its a very good solution if you insist on a single chip design, or have to work with very low supply voltages (single lithium).

 

[Wolfman29]

I don't understand - I didn't really see that on the datasheet.

So you are saying you can adjust the reference voltage of the LM3083?

 

[HIMNL9]

^ a bit of confusion was made, sorry.

First, ALL the regulators are VOLTAGE regulators (you can't directly regulate a secondary parameter as current, that is DEPENDING from voltage and resistance)

Second, i meant that, using a current as reference, instead a voltage, that regulators serie is much more efficent than usual voltage regulators, for make current drivers ..... also, when the manufacturer designed it, it was intentionally designed for work as current AND voltage limiter, not as the LM317 serie, that was originally a simple voltage regulator that they discovered it CAN ALSO work as current limiter.

Said this, yes, the LT308x serie (not LM), was originally presented as voltage regulators, that use a current, instead a voltage, as feedback reference ..... this permit you to use a direct low value resistor in serie with the load, and a high value, low power trimmer (usually 10 or 20 Kohm, 1/4W or less), for fully regulate the current from zero to max, without having to search for high power low values *EXPENSIVE* trimmers

Other than this, in current-limiter configuration, you can connect it both as source (LD with common negative), than as pseudo-sink (LD with positive common) configuration, without the need to use an add-on op-amp, and also it permit you to parallel 2 or more units for double or triple the current, if needed ..... and again, it's available in MS8E package (very small 8 pins package) for the 1A and 500mA units, and TSSOP 16 (very small 16 pins package) for the 3A type ..... basically, this mean that, using 2 of the 1A ones (or also one of the 3A ones) and some thermal transfer systems, you can build a 2A unit, regulable from 0 to 2A from a miniature trimmer, that can drop around 1,5V total, dissipate 1/4 or less than a normal LM driver, and fit inside an aixiz module

 

[Benm]

First, ALL the regulators are VOLTAGE regulators (you can't directly regulate a secondary parameter as current, that is DEPENDING from voltage and resistance)

I'd have to disagree on that really. There is no reason at all you cannot build a current regulator, its just not very common.

Obviously they always have limitations, if the load resistance is too high, they will max out at some voltage and go no further. Voltage regulators suffer from the same thing though, but only if the load resistance is too low. A voltage regulator cannot practically output an infinite current, nor can a current regulator output an infinite voltage.

 

[HIMNL9]

I'd have to disagree on that really. There is no reason at all you cannot build a current regulator, its just not very common.

Obviously they always have limitations, if the load resistance is too high, they will max out at some voltage and go no further. Voltage regulators suffer from the same thing though, but only if the load resistance is too low. A voltage regulator cannot practically output an infinite current, nor can a current regulator output an infinite voltage.

Then probably i made some other confusion (enjoy my perfect Engrisck )

Seriously, i have not said that you cannot build a current regulator, after all our drivers are current regulators ..... i've said that you cannot build a circuit that directly produce a current (cause a current is a secondary parameter, not a primary parameter, maybe i used a wrong expression before) ..... said in other words, you can produce and regulate a voltage alone (primary parameter), same as you can build and measure a resistance alone (primary parameter), but you cannot produce a current alone, cause the current that flow in a resistor is the derivation of a certain voltage applied to a certain resistance value ..... a current alone, is a thing that does not exist, without a voltage and a resistance.

So, all our drivers are, in fact, current regulators, but they do this regulating and balancing the voltage applied to the load, in the way that the flowing current remain stable also if the load change its resistance.

 

[Wolfman29]

Hmm. Makes sense. However, the Vdo is still pretty high on these - not efficient enough to beat out the dual-transistor current sink/source.

 

[BShanahan14rulz]

I used a single 7135 on a red build. Wonder if some caps could be added to make it a more viable option? Negative side current regulation, LDO, low external parts list, easy to get, cheap. 6V max, though. And we need someone with a scope to measure it. And it only increments in multiples of 350mA.

Or just the current sink mentioned above.

Should have clarified that this was a "lab-style" build, almost. Laser head was isolated from the driver. Otherwise, I'd have to modify the switch circuit board and run an extra wire

 

[Wolfman29]

How did you use a negative side current regulator for a red build, which is case negative?

 

[Wolfman29]

So I want to build one of those dual-transistor linear drivers... are there any other part numbers that I could use for the transistors? I am looking to buy locally (don't want to fork over $5 on shipping for a dollar or two on parts). However, I am looking for SMD chips (making PCBs are really easy for me now) so that it can fit inside my host... any suggestions?

Further, I read something about using a low-resistance mosfet instead of Q2 for lower dropout. Is that a good idea?

EDIT: I have some transistors on hand (unfortunately, in large packages) which are the 2N2907, 2N3906, and the 2N4403. Will any of these work as substitutes? They are all PNP's, but will they work? And does anyone know where I could salvage these parts from every day electronics? Are they common?

Thanks!

 

[Benm]

i've said that you cannot build a circuit that directly produce a current (cause a current is a secondary parameter, not a primary parameter, maybe i used a wrong expression before) ..... said in other words, you can produce and regulate a voltage alone (primary parameter), same as you can build and measure a resistance alone (primary parameter), but you cannot produce a current alone, cause the current that flow in a resistor is the derivation of a certain voltage applied to a certain resistance value ..... a current alone, is a thing that does not exist, without a voltage and a resistance.

I don't want to push it into academic discussions too far, but current actually is a base unit that is simply define as the number of electrons per unit of time. To use a water analogy: you can quite simply make a tap that spills a liter of water per minute(~current), regardless of how high that water falls from (~voltage) after leaving the tap.

Getting back to the practical part of stuff: Most general purpose PNP transistors will do for T1 in the current source, but T2 as to be able to handle the laser current. If you need some smd component ideas for the pnp version of the driver: BC857 for T1, BC869 for T2. These are NXP/Philips examples, you can obviously substitute them with transistors with similar specs from any manufacturer.