Hot Tubes' cd4049ube getting hot

[swever]

Finished an EHX hot tubes clone using gaussmarkov's layout (with the mods) and it seems to work fine. There is just one thing that worries me: the cd4049ube chip gets literally hot. At first I powered it from an 18V supply, played it for about 30 minutes and after that even the enclosure was quite warm. I opened the box and the chip was burning hot. I let it cool down and then powered with 9V. I still does get warm. Is this normal?

The current consumption of my unit is about 30mA@9V, which is a quite a bit higher than the 18mA stated on EHX's site. Does this mean there is something wrong with mine? Did not find any solder bridges or other obvious errors. Is there anything specific to check?

[swever]

Replacing the chip with another one from the same batch did not change anything. I also used the IC's from this batch to build other stuff things including red llama and I dont remember it getting hot.

[Mark Hammer]

The op-amp can handle 18V but CMOS chips should generally not be run any higher than 15VDC.

[dschwartz]

Did you ground the unised inverters?

[PRR]

CMOS logic is packaged for LOGIC. In each device pair, one is always off, current is teeny.

When you mis-use CMOS as "Linear", BOTH devices are ON.

IIRC, while CMOS takes 18V in logic operation, in Linear mode the current gets Very Large and the dissipation exceeds the package rating. (Which may be another reason we rarely see all 6 gates used linear.)

[Rob Strand]

IIRC, while CMOS takes 18V in logic operation, in Linear mode the current gets Very Large and the dissipation exceeds the package rating. (Which may be another reason we rarely see all 6 gates used linear.)

That's my memory as well.

Fig 3 here shows it should bias at half supply, but gives no indication of the drain currents.
http://www.shrubbery.net/~heas/willem/PDF/NSC/AN/AN-88.pdf

However from here,
http://educypedia.karadimov.info/library/CMOSIntro.pdf

Fig 2 shows slight asymmetry in the two halves.   For equal currents it would need an equilibrium point of 28mA and the output voltage will sit around 3.0V.

[swever]

The op-amp can handle 18V but CMOS chips should generally not be run any higher than 15VDC.

That's what I figured. But the thing is it does heat up even with 9V supply.

Also, I've had a red llama on my 18V powered pedalboard for a while and I cant recall it ever getting hot. I may have just not noticed though, will check it's chip temp tonight.

Did you ground the unised inverters?

The unused inverter stages inputs are at V+, the outputs are "in the air". There are only two unused stages.

[bool]

28mA is already a small class-ab amp idle bias territory...

[anotherjim]

Doesn't the HT use 4 inverters in linear mode? That's twice the shoot thru current as fx that only use 2 of them like that.
However, I think 2 of those inverters have pull-down resistors on the inputs that should shift the output bias slightly higher than mid-way, presumably to reduce some of the shoot thru current.

[Mark Hammer]

In 1979 or so, I acquired a clone board that a guy in the tech shop at work built.  I ran it off a 9V battery for years.  Not the same battery, obviously, but it ran fine off a battery; noting that I never had it plugged in for several hours at a time.  Never ran hot at all.  I have a clone version using a board layout RG sent me.  Works fine off a 9v battery as well, with no heat.

The original DID have an on-board transformer, but I gather that was simply because the current draw was a bit higher than what would yield the same battery life as encountered in the typical distortion pedal of the day.  It was long before 9V batteries were available in dollar stores (although Radio Shack still had their Battery-of-the-month free card at that time).

[Rob Strand]

28mA is already a small class-ab amp idle bias territory...

My apologies, I must have had a minor brain fart looking at those curves yesterday.
Those graphs are shown for Vgs=Vdd so that means each MOSFET with the gate is running at full drive.
On a normal gate you can't create this situation.  On a 4007 maybe.

The graph we are interested in is Figure 5d.
http://educypedia.karadimov.info/library/CMOSIntro.pdf

Vdd = 15V, Vin=8.5V:    ID = 7.5mA
Vdd = 10V, Vin = 6.0V:  ID = 3.4mA

The single current curve as a function of Vin without any Vdd dependence looks suspicious to me (If I had to guess I'd say it applies only to Vdd=15V).
I know when I've characterized CMOS gates at 10V I often saw maximum gain at Vin=4V.


Digging a bit further,

[Rob Strand]

n 1979 or so, I acquired a clone board that a guy in the tech shop at work built.  I ran it off a 9V battery for years.  Not the same battery, obviously, but it ran fine off a battery; noting that I never had it plugged in for several hours at a time.  Never ran hot at all.  I have a clone version using a board layout RG sent me.  Works fine off a 9v battery as well, with no heat.

From what I remember it is OK at 9V.  The situation changes at higher voltages.
For some reason the MXR Envelope Filter drops the supply voltage, perhaps to reduce power consumption.

[PRR]

> a minor brain fart

That did smell wrong.

Your fig 5d matches what I recall. Peak current for 15V is 9-11mA (depending which paper you read; also process variation because this is NOT a marketable parameter except incidentally via switching-time). This suggests 18V will make 12mA, which is 216mW. Four such lin-mode gates is nearly a Watt. The PDIP pack is rated 67C/Watt. 92 deg C is HOT. The chip is still well within the 150C limit, but is is distressing for "small audio".

If you truly need 6Vrms output, run the CMOS on 5V and run a plain opamp with gain of 3 or 4 after it.

[Rob Strand]

depending which paper you read; also process variation because this is NOT a marketable parameter except incidentally via switching-time

The papers tend to show a lot of variation.  The weird thing is, many moons ago I measured the characteristics of different brand and different era devices (Late 70's through to about 2000) and the maximum gain point agreed within +/- 0.25V.

An interesting story I got from Jim Thompson (an Analog designer who had worked for Motorola back in the 60's also later years), is he was asked to redesign the CMOS 4000 series gates.   IIRC the aim of the exercise was to use modern fabrication methods.  The main trick was to match all the timing and linear behaviours.    So inside the modern CMOS 4000 gates aren't exactly what they once were.

[amptramp]

We did many strange things in the Remote Sensing department of a company I worked at and one of them was trying to make a photosensor amplifier using a CD4049 submerged in liquid nitrogen at 77K.  Water freezes at 273.15K and typical room temperatures were 300K, so at about one quarter of the absolute temperature, we would get half the noise.  Each inverter took 8 mA with the input at Vdd/2, the halfway point where dissipation was highest.  The temperature did not rise in this setup because it just boiled away a bit more of the nitrogen.  I am not surprised you are running into temperature issues in an enclosure in air.

We concluded that we could not get proper noise performance because of the 4K of protection circuitry at the input to each gate and there was a tendency for sudden large voltage excursions which were probably the device turning tectonic at the low temperature.

[bool]

...
This suggests 18V will make 12mA, which is 216mW. Four such lin-mode gates is nearly a Watt. The PDIP pack is rated 67C/Watt. 92 deg C is HOT. The chip is still well within the 150C limit, but is is distressing for "small audio".
...

... and disstressing for the electrolytics, even if you use 105deg types.

[reddesert]

n 1979 or so, I acquired a clone board that a guy in the tech shop at work built.  I ran it off a 9V battery for years.  Not the same battery, obviously, but it ran fine off a battery; noting that I never had it plugged in for several hours at a time.  Never ran hot at all.  I have a clone version using a board layout RG sent me.  Works fine off a 9v battery as well, with no heat.

From what I remember it is OK at 9V.  The situation changes at higher voltages.
For some reason the MXR Envelope Filter drops the supply voltage, perhaps to reduce power consumption.

The MXR envelope filter is known to be sensitive to the voltage and a correct bias voltage is necessary to get the ultrasonic clock to oscillate. FWIW, I recently built one and measured an operating voltage of 4.2 V after a dropping resistor + diode (probably less than ideal, I read somewhere that 5 V or so is about right) and a current draw of only 2.3 mA, even though the circuit uses a lot of CD4069UBE inverters.

[marcelomd]

Hi,

Just my 3 cents.

1- Have you tested the chips in a breadboard? The board without the chips? Are all inverters in a chip working correctly? Sometimes digital logic burns in a conducting state, so that may explain the high current draw (and temperature). Could be an inverter that is not used in the effect.

2 - Or you may have a solder bridge somewhere, engaging one of the unused inverters.

3 - Unlikely: Interference forcing one or more inverters to be constantly switching.

I've seen all 3 happen. Sorry if you already checked them.

[Rob Strand]

The MXR envelope filter is known to be sensitive to the voltage and a correct bias voltage is necessary to get the ultrasonic clock to oscillate. FWIW, I recently built one and measured an operating voltage of 4.2 V after a dropping resistor + diode (probably less than ideal, I read somewhere that 5 V or so is about right) and a current draw of only 2.3 mA, even though the circuit uses a lot of CD4069UBE inverters.

Yes, I'm pretty sure the low Vdd voltage is to keep the power down.  Not all devices are running in linear mode, however lower voltages help power in switching circuits as well.  Vdd on my unit is about 4.0V, probably get some variation due to battery state/exact PSU voltage.

I suspect a good deal of the variability issues come from the DIY domain - wrong chips due to limited selection of parts.   MXR knew which devices work and which ones don't - maybe they also found out the hard way.   The low Vdd voltage probably doesn't help the cause.

[PRR]

... 92 deg C is HOT. The chip is still well within the 150C limit, but is is distressing for "small audio".

... and disstressing for the electrolytics, even if you use 105deg types.

Unless you strap the cap TO the chip, it won't get that hot.

You can set a cold drink on the side-table of your patio grill. It may get less-cold, but won't char.