Not thrilled with CL-60 inrush limiter in USA/160W Class A First Watt designs

Member
Joined 2011
Paid Member
I've got my First Watt M2x amplifier doing a Thermal Soak right now. M2x is a Class A amplifier that draws 160 watts from the AC mains. With USA voltage that means about 1.4 amps average mains current.

This amp has two Amphenol CL-60 inrush current limiters, connected in parallel for USA: each ICL feeds one of the two parallel primary windings on the power transformer. Thus each CL-60 sees about 0.7 amps average current. But, checking its datasheet, the CL-60 is designed to run at (1.2A <= Current <= 5.0A). Uh oh.

General Electric / Thermometrics / Amphenol don't specify the ICL body temperature on their datasheets, but other companies do. I've attached a representative one below. It runs at 149 degrees Celsius! wow.

I just measured the CL-60 ICLs in my M2x. After three hours of thermal soaking (power applied, inputs shorted), the body of the CL-60's measure 33.8 degrees Celsius using a Type-K thermocouple, and 32.9 degrees Celsius using an infrared no-contact thermometer. That's a far cry from 149 C. So I'm thinking the CL-60 may not be the best ICL for this particular amplifier.

Poking around on Mouser's website, I found the SL15-22102 Inrush Current Limiter from Antherm. Its cold resistance is 22X higher than the CL-60's cold resistance, so it gives a whole lot more protection against inrush. Its max energy rating is 40 Joules, slightly higher than the CL-60's rating of 36 Joules. Its resistance at 0.7 amps of average current is about 2.5 ohms, so the mains voltage "lost" to ICL is 1.75 volts, which is about 1.5% of the total mains voltage. I find this acceptable. I think I'll order a few of these and play around with them. Perhaps the best choice is something sort of in-between this one, and the CL-60 original.

_
 

Attachments

  • ICL_spec.png
    ICL_spec.png
    83.3 KB · Views: 2,861
Member
Joined 2011
Paid Member
Hi, do you mean that you rather would like to see the CL-60 at 149 degrees Celsius instead of 33 degrees Celsius?

I would prefer to see two CL-90s or CL-120s at 149 degrees Celsius instead of two CL-60s at 33 degrees Celsius.

However I like the other ICLs mentioned above, not made by Amphenol and with a part number that does not begin CL-. They appear to me to be even better suited than CL-90, to the M2x with USA voltages.
 
Member
Joined 2011
Paid Member
Here's the existing PCB. It's got two CL60s; I'm using the USA mains option. If you decide the two CL60s are not really in parallel, it's just that each one carries half the primary current, that's okay with me.

I prefer to drop new parts into the existing PCB rather than bodge in a different circuit topology. Modification laziness.

I also kind of like Nelson Pass's idea of getting double the max inrush energy capability, by having two $1.00 ICLs working in parallel, instead of just one.

_
 

Attachments

  • FirstWatt_PSU_schematic.jpg
    FirstWatt_PSU_schematic.jpg
    87.1 KB · Views: 3,794
Disabled Account
Joined 2002
I would prefer to see two CL-90s or CL-120s at 149 degrees Celsius instead of two CL-60s at 33 degrees Celsius.

However I like the other ICLs mentioned above, not made by Amphenol and with a part number that does not begin CL-. They appear to me to be even better suited than CL-90, to the M2x with USA voltages.

They only will reach maximum temp at startup and not after some time when they enter steady state current. Or am I thinking wrong? Inrush current limters only get hot for a short time.
 
Member
Joined 2002
Paid Member
The Amatherms are a good product they have a broad range and go up to about 750 joules if my memory is correct. They should run toasty to get the resistance down and have minimal loss. They will current hog in parallel but not an issue with dual primary windings.
 
Their lives will be significantly shorter if they run hot all the time. And if your amp is really SE class A (essentially constant power consumption from the supply) then as long as it's biasing up and voltage is holding you should be good. Inrush limiters ideally are only there on startup, then try to be wires when running (and warm).
 

PRR

Member
Joined 2003
Paid Member
...do you mean that you rather would like to see the CL-60 at 149 degrees Celsius instead of 33 degrees Celsius?

The resistance drops with heat. Cold, he's not getting "all" power expected; if they ran hot he'd have more power to run with.

I didn't find that exact part but here's the trend. (Attached)

Consider the surge-blob like it was a long extension cord. Yes, 10 Ohms means he can't surge past 12 Amps (@120V), which is long-term safe even in my 1948 house. However a lamp (or amp) on a 10-Ohm cord gets dim. If heat (self-heat or candle) brought the blob up to 150 deg its resistance would drop to 0.5r, a "better cord", which won't "dim" so much under load.

If his running current is 0.7A (each), and if R is around 7 Ohms, that's 5V of drop. 120V is already 115V. If hotted-up to "hot" it would be <1r, <1V, 120V gives 119V.

Of course Class A is all about managing the heat. If the amp was sharp-pencil designed for 115V source, "upgrading" to 119V source makes 7% more heat. We know this amp was not penny-pinched to within 7% of its life, so no death expected; still you hate to work stuff hard/hot for only 0.3dB increase of headroom.

The fact that the two lumps are on separate windings is moot. The two primaries are 99% coupled, the two lumps are "parallel" within 1%. However because the part is under-run, this is not likely a problem.
 

Attachments

  • hot!-NTC.gif
    hot!-NTC.gif
    16 KB · Views: 2,142
Cl60

They only will reach maximum temp at startup and not after some time when they enter steady state current. Or am I thinking wrong? Inrush current limters only get hot for a short time.

They get hot and that reduces their resistance, they stay hot in order to continue to pass the required current to the transformer.

I use these to soft start 1kva transformer but I don’t like them staying hot so I short circuit them with a timer relay after 10 seconds. This has the added advantage that if I have power loss for a few seconds (I live in a rural location, it happens) then they still soft start.
 
Disabled Account
Joined 2002
They get hot and that reduces their resistance, they stay hot in order to continue to pass the required current to the transformer.

FWIW, I stopped using these a long time ago but I remember it was like this:

"They get hot and that reduces their resistance, they stay warm in order to continue to pass the required current to the transformer."

I would never use parts that are supposed to be 149 degrees Celsius continuously in a consumer device.

Like you I also like the soft start principle. Always design while taking possible brownouts and blackouts into account.
 
Member
Joined 2011
Paid Member
I bought some more Inrush Current Limiter devices and ran some more tests. While taking the M2x amplifier apart to make these new test runs, I discovered a minor hook up error on my Inrush Limiter daughter board. This error (plugging the wrong wires onto 1/4" FastOn connectors) resulted in Primary#1 getting no ICL at all, and primary #2 getting both ICLs in series (!). Thus the measured ICL disc body temperatures in post#1 above are wrong, wrong, wrong. I am sure glad I decided to try some more ICLs: preparing those experiments, allowed me to discover and fix the mistake.

Although I tried numerous ICL devices in my M2x amplifier, I will only mention three of them in this summary:

  1. No inrush current limitig at all (!). ICLs removed from PCB and replaced with 16 AWG shorting bars.
  2. Amphenol CL-60, the ICL which appears on the First Watt power supply schematic drawn by Nelson Pass, and the ICL which I used initially. It is rated for 36 Joules of inrush energy, and its room temperature resistance is 10 ohms.
  3. (my preference): Ametherm SL15-60002. It is rated for 50 Joules of inrush energy, and its room temperature resistance is 60 ohms.

The power supply schematic of my M2x is shown in Figure 1 below. It's mostly a copy of the First Watt PSU, plus the addition of secondary snubbers, rail fuses, and antireversal diodes. The Inrush Current Limiters appear on the schematic as component names X1, X2. For those who might not know, M2x is a Class A amplifier that runs at a bias current of approx 1.3 amps in each of two channels.

After the inrush event is complete, the ICL resistance falls to a small number of ohms, but not zero. There is a small voltage drop across the ICL, which means the voltage applied to the transformer primary is slightly less than the full AC mains. Which means that the DC output voltage is also slightly less than it would have been without an ICL. To see this effect, I connected a DVM between the two power supply output nodes POS and NEG. Different models of ICL gave different values of total supply DC voltage (VPOS - VNEG), as shown in the table below.

ICL body temperature and supply DC voltage were measured 5 minutes after power-on, to allow the M2x Class A bias loop to stabilize and settle to a constant bias current. Temperature was measured with a type K thermocouple.

attachment.php


Replacing the ICLs with short circuits, gives the largest possible inrush current. The scope photo in Figure 3 shows this inrush current 32 amperes. That's high enough to make me nervous about blowing the 2.5 amp mains fuse, even though it's a "slow blow" type.

Installing the CL-60s reduced inrush current to 15 amperes as seen in the scope photo of Figure 4. The CL-60 discs stabilized at 59 Celsius, and the DC supplies sagged very slightly (0.2V) compared to the no-ICL case.

Changing to the SL15-60002 ICLs reduced inrush current even further; it falls to 4 amperes as seen in Figure 5. The ICL discs stabilized at 84 Celsius, and the DC supplies sagged about 1.0V compared to the no-ICL case.

MY PREFERENCE

After gathering the data, I decided that I liked the SL15-60002 better than all the other ICLs. Compared to the CL-60,

  • Its inrush current is almost 4x less
  • Supply DC voltage sags an additional 0.8V
  • Body temperature of the ICL disc is higher: 85C not 60C

I decided the benefits (point 1) outweighed the costs (points 2 and 3), and I decided I could live with the costs. So I have permanently installed SL15-60002 ICLs in my M2x.

arcane technical note: I used an on/off switch that is synchronized to the AC mains. This means every turn-on event in every experiment, occured at exactly the same phase angle of the AC mains. Randomly varying phase angle @ turnon has been eliminated as a source of possible uncerntainty.

_
 

Attachments

  • M2X_MJ_PSU.png
    M2X_MJ_PSU.png
    33.3 KB · Views: 2,785
  • TABLE.png
    TABLE.png
    8.3 KB · Views: 2,715
  • Unit_DeadShort.png
    Unit_DeadShort.png
    36.8 KB · Views: 2,003
  • Unit_Q4.png
    Unit_Q4.png
    36.8 KB · Views: 975
  • Unit_T2.png
    Unit_T2.png
    61.7 KB · Views: 1,000
Last edited:

6L6

Moderator
Joined 2010
Paid Member
Fantastic research and findings! The results are particularly telling when looking at the scope shots, the time differences are quite notable.

I'm quite surprised in that the inrush is still 15A with the CL-60s!! That the fuses generally hold is amazing, and that they blow randomly (something that's reported fairly often) is now most likely explained.

If you are interested and willing it would be interesting to measure the results of the next smaller CL-series (The CL-70 and -80) in comparison to the -60 and the SL15. The -70 and -80 are both rated for the same inrush energy as the -60 (3600 joules) so it may be an apples to apples comparison.

In the meantime, I'm going to order some SL-15-60002s. :)
 

PRR

Member
Joined 2003
Paid Member
...The scope photo in Figure 3 shows this inrush current 32 amperes. That's high enough to make me nervous about blowing the 2.5 amp mains fuse, even though it's a "slow blow" type.

It's "only" 10A peak after 3 cycles, 50mS. This part of the fuse thermal curve (rarely plotted) is practically vertical; a few 32A-10A blips won't usually blow a 2.5A fuse.

Also, it is a narrow peak. Squinting on thumb, maybe 1/10th of a half-cycle. So the heating effect of a 32A rectifier peak is more like a 10 A current; falling to 3A after the 3rd cycle.

Additional datapoint: if a 2.5A fuse blew on a dozens-mS transient, fuses would protect transistors. As we know:(, they don't.