MC1466L Power Supply

[Jwillis]

After a crushing defeat of my first attempt at creating a DIY PSU ,I decided to abandon it and try a different approach . This One is based on the MC1466L https://www.mikrocontroller.net/attachment/160010/mc1466l_tiff.pdf voltage regulator with current control by Tony Van Roon http://www.tonyvanroon.com/oldwebsite/circ/ps4002/ps4002.html. Unfortunately Tony has passed and I can't ask some questions about some modifications I've made. So far everything is going well.The voltage is a precise 0 - 40 VDC. But I haven't load tested yet.I'm hoping to get at least 4 amps  but I'm not sure if I have the power transistors balanced correctly. My math is horrible . Also I thinking the betas of Q2 may be to low to maintain a less than 0.5 mAdc at pin 5 .And my Ammeter is to low, But that can wait till version 3. What do you think of my second attempt so far.

[chris_leyson]

In your scematic you've put a short across your ammeter. RV3, R4, R5 and R6 in the original schematic are there so you can use a 0-250uA ammeter to read 0-2.5A. For now just use the 0-3A meter, you won't need RV3 and R4 and fit a 0-5A meter later on.

[0xfede]

Hi Jwillis,

you are right on the Q2 beta too low. TIP31C has an hfe of 10 at its minimum and 2n3055 20. So if you do the math 4A/(10 * 20) you will obtain 20ma which is higher than the recommended 0.5ma. You should note that there are several hfe grades for TIP31C and even the higher specced device has a beta of 40 which yelds 5ma of load on pin 5.

Since you have 32Vac input you should expect 32*sqrt(2) in DC and that yelds ~45VDC across C1. If you short the output with this configuration the 4 pass transistor will dissipate 45*4 = 180W that is a huge amount of power. That's not healty and you should consider a multitap transformer.

In your design there is a short circuit from IC1 pin11 and PSU output that short both the ammeter and the current error amplifier. This will mean that your circuit will not regulate the current and your ammeter will not measure anything.

It is also strongly recommended to add bleeding resistors across C1 & C2 and across C3.

I recommend to change D1, D2, D3 with faster ones (like 1n4148).

The diode D7 should be at least rated for the output current (4A) where the 1n4007 is only 1A.

I would also recommend an output on/off system.

Hope that helps.

Best,
0xfede

[Benta]

My question is how you even got hold of an MC1466L? It's obsolete since 40 years.

[David Hess]

My question is how you even got hold of an MC1466L? It's obsolete since 40 years.

That is the problem all right.  The MC1466 did not really go obsolete since the 723 did not but it never became popular enough to survive.

The functions of the MC1466L can be replaced with a pair of 723s which is reasonable given their low cost and there is a power supply schematic out there which does this.

[Wimberleytech]

My question is how you even got hold of an MC1466L? It's obsolete since 40 years.

Looks like they are on ebay with a date code as recent as '93
https://www.ebay.com/itm/1pcs-100-New-And-Original-MC1466L-Integrated-Circuit-CDIP-14/181609137974?_trkparms=aid%3D222007%26algo%3DSIM.MBE%26ao%3D1%26asc%3D20160908110712%26meid%3Db9f22936153e4dd8b5670e59ff5b595a%26pid%3D100677%26rk%3D14%26rkt%3D21%26sd%3D401446563631%26itm%3D181609137974&_trksid=p2385738.c100677.m4598

[floobydust]

OP, as far as your design mods, there are some improvements I would suggest, if you want to stick to the MC1466 circuit.

Reverse-engineering and making a MC1466 from discretes. It was done to save a Lamda PSU; OSH Park pcb available it looks like. http://tubetime.us/index.php/2014/09/28/a-miniaturized-discrete-mc1466/

I'd go with a couple op-amp design, as you have a second power transformer for them, it makes it easier.

[Jwillis]

In your design there is a short circuit from IC1 pin11 and PSU output that short both the ammeter and the current error amplifier. This will mean that your circuit will not regulate the current and your ammeter will not measure anything.

I suppose I could just shunt the ammeter of the output.That would be an easy rewire.
I'm confused about pin 11. Isn't pin 11 current sense and pin 10 is current control ?I have a hard time reading the block diagram on the data sheet.

My question is how you even got hold of an MC1466L? It's obsolete since 40 years.

I have a time machine. That's what this PSU is for so I can boost the Power and go back farther than 40 years.
Seriously I picked up a bunch of them fairly cheap on EBay . The MC1566L is the really hard one to find.

[David Hess]

OP, as far as your design mods, there are some improvements I would suggest, if you want to stick to the MC1466 circuit.

Reverse-engineering and making a MC1466 from discretes. It was done to save a Lamda PSU; OSH Park pcb available it looks like. http://tubetime.us/index.php/2014/09/28/a-miniaturized-discrete-mc1466/

A pair of 723s could also be used and you can buy a lot of new 723s for the price of one NOS (new old stock) MC1466.

I'd go with a couple op-amp design, as you have a second power transformer for them, it makes it easier.

The distinction between a MC1466 or dual 723 design and a traditional operational amplifier design is that the former provide access to the transconductance output node which is used for frequency compensation.  This point can be placed in parallel through diodes with another transconductance error amplifier output to allow fast switching between constant voltage and constant current regulation.  Almost all operational amplifiers do not provide access to this point slowing switching time between constant voltage and constant current regulation.  The best that can be done for them is to clamp their output through feedback somehow or just select an operational amplifier with high slew rate.

Some old operational amplifiers which support external compensation can be used the same way like the 301A but taking advantage of this means not being able to use a general purpose operational amplifier.

[Jwillis]

I read another site along the same lines using through hole discretes .http://chemelec.com/Projects/MC1466/MC1466.htm. Fun Fun Fun.
This is what I used to configure the power transistors on mine.I just need to change the TIP31C to something more suitable.

The thing that put me off the LM723 is I can't seem to find how to pull the voltage to zero.Must likely I'm not reading the data sheets correctly.

[floobydust]

I've never seen as double LM723 design, but sadly notice the IC is end-of-life at Digikey and Mouser
Wait, is this circuit you mean: https://www.eevblog.com/forum/beginners/bangood-psu-enhancements/msg1347326/#msg1347326


For CC/CV "wired-OR" if I correctly understand your use of the LM723 compensation pin, I use a couple diodes from the output of separate CC and CV op-amps.
It's still fast, but most of these linear power supply designs have terrible overshoot and sluggish pass-transistor stages, like OP's which is missing E-B resistors, so they will switch off slow.

Too bad we couldn't fix that chinese 0-30V 3A kit design into something robust and accurate. I ended up tossing it, it was beyond frustrating. It tortures the Internet.
https://www.eevblog.com/forum/beginners/bangood-psu-enhancements/msg897523/#msg897523
https://www.eevblog.com/forum/beginners/0-30v-2ma-3a-dc-ps-kit-messed-up!/
http://www.qsl.net/z33t/dc_0-30v_0-3A_eng.html

[0xfede]

In your design there is a short circuit from IC1 pin11 and PSU output that short both the ammeter and the current error amplifier. This will mean that your circuit will not regulate the current and your ammeter will not measure anything.

I suppose I could just shunt the ammeter of the output.That would be an easy rewire.
I'm confused about pin 11. Isn't pin 11 current sense and pin 10 is current control ?I have a hard time reading the block diagram on the data sheet.
....

....
[/quote]

Since you have shorted pin 11 and pin 9 the differential voltage between these 2 pins will be always 0 and therefore no current regulation is provided. In the datasheet at the bottom of page 10 you can find a schematic with a shunt resistor between pin 11 and 9. For semplicity you can use the same shunt resistor for your ammeter.

I read another site along the same lines using through hole discretes .http://chemelec.com/Projects/MC1466/MC1466.htm. Fun Fun Fun.
This is what I used to configure the power transistors on mine.I just need to change the TIP31C to something more suitable.

It is much easier if you add another small transistor before TIP31C. Try doing the math.

The thing that put me off the LM723 is I can't seem to find how to pull the voltage to zero.Must likely I'm not reading the data sheets correctly.

Everyone who started learning electronics combined toghether all the parts scrapped from everywhere without exception. I remember that it was real fun and it made me grow.
You are doing well and should continue on this track and you should not mind about all the comments that try to hijack your nice PSU project.

Best,
0xfede

[Jwillis]

OH I see that now.Thank you 0xfede. That makes sense now.I should probably put that diode back to for short protection as well .I have some darlintons with better HFE on the way to fix the TIP31C issue. I don't have  any 4 amp diodes but do have 10 amp ones. It's over kill but should work ok '
I don't hijacking and constructive criticism as long as it includes good clear information. I appreciate the help from everyone thank you so much.

[Neomys Sapiens]

Is it one transformer of two that you are using? I often try to include a circuit that either
- clamps the voltage setting to zero/lowest
or
- switches the main transformer in only
if and when the auxiliary voltage(s) is/are present and stable.
Prevents ugly transients with some regulator circuits. I would have to look into the 1466 appnote that I have somewhere to see if it applies.

[Jwillis]

I'm using 2 separate transformers since I didn't have a multi wind with the right voltages .They are wired at the primary in parallel to switch on at the same time .The MC1466L requires 2 isolated sources. The auxiliary can't can't be less than 20VDC or the IC won't work and can't exceed 30VDc (MC1466L) 35VDC (MC1566L) or it will burn out. The Main transformer is a multi tap primary 0v,100v,116,135v with 15v-0-15v and 24v secondaries .I'm using the 116v tap.The 24v isolated secondary is to high for the MC1466L but not the MC1566L, which I could not find at the time . The Aux transformer Is a 115v primary with 9v-0-9v secondary for 18v.

[floobydust]

I read another site... {stuff deleted} http://chemelec.com/Projects/MC1466/MC1466.htm...

From that site, this schematic - so add the missing shunt-resistor RS from pin 9-11. The four-transistor emitter follower is too much gain, I would expect it to oscillate badly and be very slow as it's missing E-B resistor on the first transistor. Using a triple emitter follower with E-B resistors (combining the two options) I think would be better.
I wonder if the voltage-sense transistor Q16 makes the voltage setpoint drift with temperature?
Some MC1466 circuits have a clamp zener between 7-9. I wasn't sure why. It might be to prevent the IC from getting damaged.

[David Hess]

I've never seen as double LM723 design, but sadly notice the IC is end-of-life at Digikey and Mouser
Wait, is this circuit you mean: https://www.eevblog.com/forum/beginners/bangood-psu-enhancements/msg1347326/#msg1347326

That is the one I remember but the MC1466L design could be duplicated with a pair of 723s by using their compensation pins as outputs.

For CC/CV "wired-OR" if I correctly understand your use of the LM723 compensation pin, I use a couple diodes from the output of separate CC and CV op-amps.
It's still fast, but most of these linear power supply designs have terrible overshoot and sluggish pass-transistor stages, like OP's which is missing E-B resistors, so they will switch off slow.

The advantage of wired-OR with the transconductance outputs is that there is no integrator windup like there would be with a pair of normal operational amplifiers where charge has to be removed from the internal and external frequency compensation capacitors when switching modes.  This is more important when the output stage is fast.  Designs which ignore this use more output capacitance to prevent overshoot.

[Jwillis]

I decided to just swap out the TIP31C for a TIP100.They happen to be handy.Anyway the TIP100 has a Hfe (Beta1)of 1000.Here's my problem when I calculate ,since the 2N3055,s are in parallel ,do I calculate for the average Hfe (beta)for the 4  2N3055's  or add them together. I know that in series they're multiplied but can't find information on the beta of parallel transistors.

Imax/B1*B2+B1+B2+1<=0.5mVDC

[David Hess]

Since they are in parallel, average them together.  In practice when I use paralleled power transistors, I grade them for matched Vbe which also matches the hfe so the average is about the same as any one part.

[joeqsmith]

I believe I designed this power supply around that same part.   

[Jwillis]

Nice! I like the interface.I've ordered some MC1566L 's and considering toroid transformers for the final version .Also I'm going to add regulation at the Aux ,mode indication and short protection with indication. Might even add a 5V USB just the the heck of it.

[Neomys Sapiens]

Nice! I like the interface.I've ordered some MC1566L 's and considering toroid transformers for the final version .Also I'm going to add regulation at the Aux ,mode indication and short protection with indication. Might even add a 5V USB just the the heck of it.

Yes, I prefer to do so too. One of my earliest LabPS was a mod of a discrete ITT application where I fitted a 3-transistor negative regulator to the aux voltage - it helps with drift, noise and tolerances.

On the topic of mode indication/short protection: I like those circuits where you can select from the classic LPS action (CV/CC) or a electronic fuse action, that trips on limit and needs an explicite reset.
Beyond this I have thought about integrating a current peak indicator mode.
Playing around with such add-ons becomes easier when one normalizes the measurement signal to 0..10V or other convenient range.
Also, I would add a thermal protection for the transformer and the power transistors. This could be done by thermoswitch, but thermistors would make it possible to have a dual warning/cutoff action.

 

[Jwillis]

I figured since it applies to this project I would just ask here instead of making a new post .
Here,s the problem .The smaller transformer failed on the low side prematurely .Initially I thought it may be just a poor transformer since I'm not drawing enough on it to be a factor.I'm only drawing about a third of its rated output.
Then I remembered reading something about toroid inductors causing magnetic interference with each other if not positioned correctly when in close proximity .
Is it possible that because of the proximity with each other.about 2 inches, that the larger main E transformer caused interference with the smaller auxiliary E transformer  causing it to fail.Before it gets replaced I'm wondering if a metal shield between each one would help , orienting them differently  would work , making sure that they aren't out of phase with each other  or just should I just move everything to a larger chassis.
Or am I just over thinking this and it was ,although brand new,just a bad transformer.   

[floobydust]

With toroidal transformers, you don't want their mounting hardware to make a shorted-turn.
Usually it's a rubber washer, long bolt and metal washer. Nothing that would make a complete loop through the center. 2" apart seems fine, stray coupling should not cause a failure.
Unless you got the primary or secondary phasing wrong? They sometimes have multiple windings and mixing that up makes smoke.

[Jwillis]

I'm probably over thinking this.Both are off the shelf E core transformers so I don't think they would have  out off phase winding's . I had them orientated 90 degrees with each other .I have a couple news with higher ratings coming today.It may be possible that that it was just a bad one or it was over rated .I'll have to pull it apart and measure the core and secondary winding.I'll rewind it later.