The crowbar doesnt work without a fuse. Bonus tip, you can place an RC filter on the input, because the low currents involved. I would place some elco-s also on it, because they are nice for filtering noise. Also, you might want to have a TVS on the output, because ESD could drift your device, and you dont even notice it. I will try to find and send my ghetto 10V reference schematic.
Ah, yeah, I wasn't sure about output protection. I'll add that for sure.
For the RC network on the input, you mean at the battery terminals? Or between the 15V regulator and LT1236?
I'm not sure what you mean by using a fuse? The transistors are just supposed to illuminate the green or red LED, but they arn't switching between green and red cleanly, more fading between the two colours.
would a much cheaper ±25ppm/°C resistor be ok here seeing as the trimpot is ±100ppm/°C anyway?
Link to BOM.
https://www.digikey.com/BOM/Create/CreateSharedBom?bomId=7607097
Hello,
- The TC of the trimming scheme is reduced at least by a factor of 70. So a 25 ppm/K resistor is sufficient against the (typical) 3 ppm/K of the A-version of the reference.
- I fear the BOM is not visible to all (at least not for me).
- I would add a 100nF capacitor at the output against EMI (eventually in parallel with 10uF Ta). Or at least 1nF if you need fast transient regulation.
- Since the voltage regulator (which one?) has variable power dissipation it could be located more far from the reference. (But then you will need another 100nF at the LT1236 input.
- The plastic package will be influenced by humidity. My LT1236-5 are influenced around 0.5ppm/% rH with a time constant of 3-5 days.
with best regards
Andreas
Apparently the class-AB output stage does not benefit from a preload but the output could have a snubber, from the datasheet:
In critical applications, a 10 microfarad solid tantalum capacitor with several ohms in series provides optimum output bypass.
I would replace the small ceramic bypass capacitors with the same 10 microfarad solid tantalum capacitor used above; just make them all 25 or higher.
I would replace the small ceramic bypass capacitors with the same 10 microfarad solid tantalum capacitor used above; just make them all 25 or higher.
Hello,
the tantalum is for settling time of the LT1236. (below 1 MHz)
And usually 10uF have around 3-10 Ohms series resistance. (ESR)
The 100nF (low inductance foil or ceramics) is against EMI. (1 - 100 MHz and partly above)
So you will need both. (if settling time is critical)
with best regards
Andreas
Ah, yeah, I wasn't sure about output protection. I'll add that for sure.
For the RC network on the input, you mean at the battery terminals? Or between the 15V regulator and LT1236?
I'm not sure what you mean by using a fuse? The transistors are just supposed to illuminate the green or red LED, but they arn't switching between green and red cleanly, more fading between the two colours.
Well, I guess most of us understand the zener / transistors on / low batt indicator. But for a clean switching you should use an op amp / comparator. For whistles and bells and safety belts on a LT1236 or any similar device, watch this:
Alrighty, Here's the latest revision. I haven't played with the low battery circuit yet. Let me know if I missed something...
I think I messed up and D1 and R7 need to be swapped.
Changes:
- Added the TVS diode across the output (It's a SOT23 package with two diodes inside, so I just ganged them up).
- Added the 10uF Tant (1.6 ohm ESR) and the 5.1ohm resistor across the output. (Is there any way to optimize that resistor?)
- Added 0.1uF cap across the output.
- Routed the ground trace from the LT1236 directly to the output terminal as shown in the datasheet, page 8.
Ah, yeah, I wasn't sure about output protection. I'll add that for sure.
For the RC network on the input, you mean at the battery terminals? Or between the 15V regulator and LT1236?
Either is an option, you have a lot of overhead for the voltage.
I'm not sure what you mean by using a fuse? The transistors are just supposed to illuminate the green or red LED, but they arn't switching between green and red cleanly, more fading between the two colours.
It looked like a overvoltage protection, but it is not. Just ignore what I wrote about this.
Looks like your TVS diodes on output are backwards. And D1 R7 look correct.
Other suggestions- Adding hysteresis to your under-voltage indicator may help. Add a high value resistor (guessing 1M) from the collector of Q2 back to the base of Q1. Lowering the value will give it more 'snap' but will alter your trip points slightly.
Nice ground plane! I would add several vias to connect top and bottom planes together throughout the board; especially on both sides of long traces that tend to break up a given plane.
I second the suggestion of moving the 15V regulator away from reference. Perhaps locate it by power switch. Then add RC filter between C2 and U2 physically located near U2.
Add 2-10 Ohms or so in series with SW1 to prevent a resonant ring-up of voltage at C1 due to parasitic inductance. Alternatively, add a cheap electrolytic in parallel with C1; the ESR will provide the damping.
Nice enclosure. Who's the source?
Yep, I saw the TVS diodes after I posted...
Derp.
I already swapped D1 and R7 in the latest schematic, so they should be right now. I have them reversed on the first revision/PCB.
I have added a spattering of vias too, the more vias the betterer.
I'll move that vreg as suggested, for the RC filter, what values etc would you suggest?
Alrighty, I had another poke at things and here's where it's at.
I moved the voltage regulator to the other side of the PCB, over near the power switch.
Added 10uf cap to the output of the vreg.
Added 0.1uf cap near the voltage reference.
Added input protection stuffs.
Added some vias here and there.
Ran the GND trace from the voltage reference directly to the output jack.
Let me know what you guys think.
Finally made some more progress.
I think I've nailed the battery monitor circuit. Took a complete redesign of that section, but it works well. (The red highlighted area).
With the chosen resistors, it has a tiny bit of overlap of the led's as they transition from green to red, so there is a bit of orange just above 16.1V, so the LED is solid red right at 16.1V, just at the Vregs dropout voltage.
Let me know if you think I can simplify it any more, it works well as is, but if I can reduce parts count, it makes building it simpler and cheaper.
The power supervisor still looks pretty complicated and high power.
Another option is to use an TL431 with a divider - it would make the LED lid above the set limit.
The normally a 9 V battery is supposed to be empty at less than about 7-7.5 V. If rechargeable ones should work well it may be even a little lower. A suitable regulator would be 12 V low drop, so that anything above about 12 V is acceptable, one could get a simple LED signal just from an transistor switched from the in/out difference. The LT1236 needs enough current to drive the base current.
Some LDOs like LP2951 even include an power good / error signal - so not extra voltage supervision needed. The LP2951 would need a divider to set the output to around 12 V. There maybe ready made 12 V versions too - though not sure about that.
Having another look at the datasheet for the LT1236, I reckon I might be able to ditch the voltage regulator all together... A feasible idea?
And if I replace the LM358 with a cheaper TL431, and raise the values of the voltage dividers to reduce wasted power, that might be a reasonable solution?
The LT1236 still has some effect from a variable supply. Not much (e.g. 1 ppm/V range), but still avoidable.
The extra regulator could also provide some protection from spikes and possible current limiting in case of latch-up. Also less power loss from self heating when loaded a little more.
A higher impedance divider and TL431 is very feasible. Modern LEDs could work with around 1 mA or less.
So my favorite would be the LP2951 set to about 12.4 V and an LED from the error signal.
Or bootstrap the LT1236 supply from its output using an operational amplifier.