Use of Zetex FCX617 and FCX717 in Linear Regulator

[Kercheval]

I have recently been working with linear regulators for Luxeon 1W and 5W parts and have been having problems with the use of the Zetex FCX717 and Zetex FCX617.

Working with both circuits described here, if I use a darlington style transistor like the TIP120 rather than the FCX617, these circuits perform as expected. If I use the FCX617, then these circuits do not limit current (the LM334 circuit uses the FCX717 PNP, but has the same issue).

After I noticed a problem in both circuits I decided to use a simple test case.

For testing, I used a simple 2 transistor current limiter (a simple version shown in the same thread above). This circuit does a reasonable job across 5V <= Vin <= 15V when using the TIP120. For larger currents, R1 across Vin and the TIP120 base should be larger to flatten the output current curve, but other than that variable, Rsense is the only variable part (to determine current amount). When the FCX617 is used in place of the TIP120 (no other modifications), the FCX617 seems to act simply as a switching transistor and the circuit does not act as a current limiter at all.

I have verified proper operation of all components and have used multiple new parts for both transistors so that I am certain it is not one bad component.

I do know that some of you have had success using this part and it has a very low Vce which I would like to take advantage of, but I am having NO success to date using this transistor as anything but a switch.

Can anyone offer an explanation of this behavior? All feedback would be appreciated.

**UPDATE**

Modified to give results of this circuit.

 

[Doug S]

[ QUOTE ]
Kercheval said:
I have recently been working with linear regulators for Luxeon 1W and 5W parts and have been having problems with the use of the Zetex FCX717 and Zetex FCX617.

Working with both circuits described here, if I use a darlington style transistor like the TIP120 rather than the FCX617, these circuits perform as expected. If I use the FCX617, then these circuits do not limit current (the LM334 circuit uses the FCX717 PNP, but has the same issue).

After I noticed a problem in both circuits I decided to use a simple test case.

For testing, I used a simple 2 transistor current limiter (a simple version shown in the same thread above). This circuit does a reasonable job across 5V <= Vin <= 15V when using the TIP120. For larger currents, R1 across Vin and the TIP120 base should be larger to flatten the output current curve, but other than that variable, Rsense is the only variable part (to determine current amount). When the FCX617 is used in place of the TIP120 (no other modifications), the FCX617 seems to act simply as a switching transistor and the circuit does not act as a current limiter at all.

I have verified proper operation of all components and have used multiple new parts for both transistors so that I am certain it is not one bad component.

I do know that some of you have had success using this part and it has a very low Vce which I would like to take advantage of, but I am having NO success to date using this transistor as anything but a switch.

Can anyone offer an explanation of this behavior? All feedback would be appreciated.

[/ QUOTE ]

John,

Actually a number of circuits were discussed in that thread. Are you talking about the two PNP circuit per MrAl posted on 12/31 2:54am and the LM334 as per MrAl posted 12/29 7:37am but as modified per my suggestion on 12/13 8:04pm [added resistor to + terminal of LM334 plus specific component values. Please give the specific component values you are using, your intended design current, and your actual observed current.

 

[Kercheval]

[ QUOTE ]
Doug S said:
Actually a number of circuits were discussed in that thread. Are you talking about the two PNP circuit per MrAl posted on 12/31 2:54am and the LM334 as per MrAl posted 12/29 7:37am but as modified per my suggestion on 12/13 8:04pm [added resistor to + terminal of LM334 plus specific component values. Please give the specific component values you are using, your intended design current, and your actual observed current.

[/ QUOTE ]

/ubbthreads/images/graemlins/smile.gif Somehow, I thought you might ask those questions.

I actually tried all three of the circuits mentioned in that thread with the same result for all of them. My specific test case was the two NPN transistor circuit.

Here is the diagram (lifted from the thread) for the test setup.
<font class="small">Code:</font><hr /><pre>
+ O-+- +LS- ----+
| |
|------- |
> |
R1 < |
> |
+--< Q2
| |e
Q1 >--+
e| |
| >
| < R2
| <
- O---------+---+
</pre><hr />
Current for this should be I = Vbe/R2 (Vbe of Q2) The components in use are

Q1 = PN2222A
Q2 = TIP120 (Vbe ~.7) or FCX617 (Vbe ~.8)
R1 = 1K Ohm (Also tried 470 and 2K)
R2 = 1 Ohm (2W 5% Metal Film)

The regulation patterns (rough measurements) are
<font class="small">Code:</font><hr /><pre>
TIP120 FCX617
Vin I(mA) Vls I(mA) Vls
-------------------------------------------
1 0 .9 1 1
2 0 1.7 2 2
3 2 2.5 3 3
4 2 3.5 4 4
5 4 4.5 17 4.9
6 59 5.3 175 5.75
7 271 6.0 454 6.2
8 574 6.40 925 6.6
9 688 6.45 >1.5A
10 712 6.47 Not gonna do it...
11 715 6.48 Yeah, right...
</pre><hr />
Where Vls is the voltage measured across the 5W Luxeon

The single difference in these measurements was swapping the FCX617 in place of the TIP120. I tried two different new FCX617s to ensure no damage to the component and checked the connection points several times (the TIP120 regulation worked so this was correct, I believe).

 

[MrAl]

Hello there Kercheval,

The gain of the 617 is very high and the be drop might
be smaller then usual. To compensate for this, you
might try either or both of these:
1. go higher with R1 (say 5k)
2. add a resistor from base to emitter (on the 617)
start with maybe 1k.

Note a darlington will have 2*be voltage drop so that
could be why it would appear to work better.

Try this and let us know how you make out.

Good luck,
Al

 

[Kercheval]

For the test circuit, I found the issue.

Something was not adding up...

hFE tested for the 617 is 550, but the current limiter circuit I am using should be fine with a high gain. Hmmm. OK, using a pot, I tried R1 all the way up to 50K without ANY change, then I remove R1 and current was still flowing! Huh??? There was no resistance between C and E on the 617! On TWO seperate parts? Heat or static, I don't know. /ubbthreads/images/graemlins/blush.gif Carefully, I prepared a THIRD 617 and validated hFE and Vbe. Now the circuit is working.

RE: The other two circuits: the two 717s I used are working so that does not explain the problem there. I will assume user error and place it back together on the board. Bit by bit by bit...

I may post the results of the various circuits I am putting together with regulation charts if anyone is interested.

Thanks for the responses.

 

[Kercheval]

Results of testing for this circuit

Here is a bit better schematic of this circuit.

Here are the regulation curves I measured for the TIP120.

Here are the regulation curves I measured for the Zetex FCX617.

Notice how much less overhead the FCX617 circuit takes in the regulation of the current here.

Here is where you can find all of the details about the circuits as I put them together.

Finally, here is where I will be placing all of my results as I explore the wacky world of regulators for LED circuits. /ubbthreads/images/graemlins/crazy.gif

Thanks again to the experts here.

 

[Doug S]

Re: Results of testing for this circuit

John, it's good to hear that you got it working. My apologies for confusing the issue by referring to Al's circuit as a 2 PNP version. His diagram is like yours, using 2 NPNs. This circuit can of course be implemented with NPN or PNP with suitable rearrangement of the circuit. I was careless in looking at Al's diagram and was seeing it as PNP for no better reason than that I habitually use the PNP version for this circuit. The PNP and NPN versions are equally functional. The LM334 circuit, however, is only suitable for PNP.

 

[MrAl]

Re: Results of testing for this circuit

Hello again,

Im glad you found the problem, and that problem wasnt
related to the 617.

I forgot to mention that i noticed that the regulation
with the 617 is sharper then with a lower gain unit.

The curves you posted are very nice too. What i would
really like to see next is a curve of output current vs
temperature for a temperature range of say zero degrees
C to maybe 50 or so.

Also, did you try to optimize the base resistor so you can
get down lower in input voltage, like to 3.5 volts maybe
and still get regulation?

Take care for now,
Al

 

[MrAl]

Re: Results of testing for this circuit

Hello again,

I tried several values of resistors and found that
when using the Zetex 617 transistor 1k seems to
be good for the base drive resistor in this circuit,
as that doesnt take the efficiency down too much
but still allows you to maintain regulation down
to about 3.6v or so.

Good luck with your LED circuits,
Al

 

[Kercheval]

Re: Results of testing for this circuit

The regulation curves above were using R1=1K so this is good news. The curve was using a 5W Luxeon with a Vf of 6.45 at 700mA.

MrAl, for reference, what load are you driving here?

Thanks for the followup, MrAl.

 

[MrAl]

Re: Results of testing for this circuit

Hello again,

I guess i should have mentioned that i was running a
simulation using a 1w LS which dropped close to 3.2 volts.
The sense resistor was 1.7 ohms. The current was still
about 300ma at 3.7 volts input, so this means it was
actually out of regulation at that point, even though
there was still significant current flowing.

The fact that it was still putting out significant current
at 3.7v input is good, but still not as good as my other
circuit which maintains regulation right down to 50mv
above the LS's voltage drop, which comes out to about
3.3 volts input.
The reason for the difference (3.7v vs 3.3v) is only
because of the value of the sense resistor, which has
to be 2 ohms for the 1w LS and 1 ohm for the 5w LS, but
with the new circuit the sense resistor comes down to
a mere 0.1 ohms, which is 10 to 20 times less then with
this two transistor circuit.
Driving the 5w LS with the new circuit will require a little
more then 80mv above the LS's voltage unless you are willing
to use a 0.05 ohm sense resistor.

I should be able to post this circuit very soon now.
It uses one op amp and one transistor and a small number
of resistors that can be SM type.

Good luck with your LED circuits,
Al

 

[Kercheval]

Re: Results of testing for this circuit

I look forward to seeing your new circuit.

Thanks for the clarifications.

jbk

 

[moraino]

Re: Results of testing for this circuit

Hi, thanks for sharing.

It looks like 8 rechargeable batteries will work fine here.

Is there any idea of roughly what kind of effeciency we are looking for?

Henry

 

[Kercheval]

Re: Results of testing for this circuit

Well, since I am currently looking at linear circuits, I have been measuring only Isource (mA at supply) and not Iload (mA at load). Since there is no power conversion in a linear circuit, these two values are approximately the same (minus small leakage currents through the control transistor).

The efficiency of most linear designs are pretty well known. For this particular circuit, there is a .4-.6V drop across the regulator (using the FCX617). To maximize efficiency, you try to minimize the drop out voltage. When this is brought down to around .1V (or less) with a small part count, then we have a highly reliable supply which, for Vin approximately Vout, the efficiency would be in the high 90s.

The particular characteristics of these various designs is interesting to me (a bit geeky I know) and I am of the 'simpler is better' school of thought when it comes to electronics.

/ubbthreads/images/graemlins/smile.gif More than you were asking, I know...

 

[MrAl]

Re: Results of testing for this circuit

Hello there again,

I've posted the new circuit in another thread:
"New Constant Current Circuit..."

This is one of the best approaches i have found
so far for LED drivers of the linear type, and
as you already know sometimes it beats or equals
a switching type regulator.

Good luck with your LED circuits,
Al

 

[Kercheval]

Re: Results of testing for this circuit

I have create a new circuit using an opamp to drive the power transistor. It is posted at http://www.candlepowerforums.com/ubbthreads/showflat.php?Cat=&Board=UBB6&Number=199576&page=0&view=collapsed&sb=5&o=7&fpart=1

 

[MrAl]

Re: Results of testing for this circuit

Hello there Kercheval,

I had run into the LM10 a while back when i was first looking
into linear regulators of this type. I had rejected it
because of two reasons:

1. Price, a little high
2. Frequency response, about 10 times lower then the LM358
and similar National parts with built in reference.

Since the frequency response is about 10 times lower then
the LM358, the surge current will most likely be much
higher and last much longer then with faster (1MHz or better)
chips.

There are probably more modern parts too that do the
same thing as the LM10 but faster :-)


Good luck with your LED circuits,
Al