7 channel Low side driver .. TPL7407L/ULN2003 replacement

[Jadeit]

I need to control 6 solenoids 5W 12 or 24V each. A maximum of 3 can be connected at the same time.
So far we have only used a 24V solenoid so I used the TPL7407L
Unfortunately, 12V solenoids require twice that current and the TPL7407L cannot be used.
Is there any transistor array with lower RdsON transistors?
So far, I designed an AO3422 + 2 resistors and 1 diode for each channel, but instead of one IC, it's 24 components.

Thanks

 

[KlausST]

Hi,

I can´t follow your worries. Show your math. Refer to the datasheet.

Klaus

 

[D.A.(Tony)Stewart]

I think the TPL7407LA is your best choice but with a 5~10W heatsink on top. Nothing else comes close in arrays.

 

[BradtheRad]

It should be feasible for multiple transistors to share a greater Ampere level. Use N-device or P-device as dictated by polarities in your system.

In case of mismatched components adjust each bias so that no transistor is overburdened. In this simulation beta values don't match yet every resistor & transistor stays under 60mW.

 

[D.A.(Tony)Stewart]

What are your driver source specs? 3.3V 5V? mA max? Max ambient temp?
long term qty ?

 

[BradtheRad]

Hmm... Your data-sheet link to TPL7407L speaks of paralleling the outputs in case one is insufficient.

Therefore by joining two outputs together it sinks enough current to pull in one solenoid at 500-1000 mA. To handle 6 solenoids requires two TPL7407L.

 

[D.A.(Tony)Stewart]

If this were high volume and you wanted to save a buck per array, this would cost 10 cents.

--- Updated Dec 10, 2023 ---

if you are doing low volume, this is more economical in 12V/24v.
Add 1N4148's to relay to absorb arc on opening.

BTW the NMOS switches are 3.25 ohms max and the PN2222A is 2 Ohms max but the Darlingtons add an extra diode drop use more power which limits the use of 500 mA /ch and needs a bigger heatsink. So if DIY, you still may need as heatsink on the TPL7407 in high ambient.

 

[Jadeit]

Hi,

I can´t follow your worries. Show your math. Refer to the datasheet.

Klaus

OK 5W load on 12V is current 420mA, TI in the TPL7407L datasheet does not list the Rdson transistor in the IC, it uses marketing words like " Power Efficient (Very low VOL)". According to the Figure 1 is Rdson on 70C 2Ohm. For current 420mA is power dissipation 350mW, for 3 active transistor I have 1W power dissipation on SOIC 16 .
According to Figure 5 in this, they can absorb only 800mW without cooling at 40C. Yes, I can stick a heatsing on the SOIC16.
However, the AO3422 I mentioned has a Rdson max of 0.2 ohms, it is in a SOT23 case and there are 100 pcs of them for a one dollar. So I thought, surely someone makes a replacement UL2003 with similar transistors, because they would be useful for controlling solenoids and would be easy to sell.
My discreet solution
minimalistic solution, I couldn't cut it any further
6xAO3422
6x diode maybe MMBD914, It's only for 200mA, but it can withstand 1A for 1s and I don't open the solenoids in quick succession
2x resistor array 4x4,7k Protects the MCU when the transistor is destroyed by limiting the current to the GPIO to 5mA
Yes, it would still require a resistor between G and GND defining the level before initializing the MCU, but in this case it probably doesn't matter, the solenoids are slow.

 

[KlausST]

Hi,

My datasheet says:
* Max current per channel: 600mA,
* Continous current per Channel 500mA
* total GND current 2A max
so why are you worried about 420mA? and you said you may use 24V solenoids, then the current is just 210mA.
I don´t get why you say that using TPL7402 can not be used.

My datasheet Figure1 states: V_OL of about 0.8V at 420mA and 70°C.

So your calculation of 350mW per channel is correct. Yes, if you design it to be difficult. But why make it hard?

Let´s just assume - like you said - you use 24V solenoids. --> half current --> 1/4 of power dissipation.
so from 350mW --> 90mW.

And as Brad said: you could parallel several channels. If you just use 3 solenoids per IC (2 ON, 1 OFF), you get
90mW max per IC.

***
But for sure there are other ICs: (quick search on Farnell)
* MPS6606
* TPD2017
* TPD2015
* L9733
* BD8LA700
* BD8LB600
* L9822
* LE7230
* L9825
Many of them have additional features, like: serial interface, current limit, voltage limit, free wheeling diodes, overtemperature shutdown, diaognostics ...

Klaus

 

[D.A.(Tony)Stewart]

Sharing ports does not reduce the total power dissipation in the package. You must give more specific design limits before you can clearly choose from all the options given to you. I think a conservative max Tjcn at 150 to 200'C rise /W will be a gating item along with total costs.

 

[dick_freebird]

Do you have something against a bunch of small outline logic-level
MOSFETs and a couple of 74HC octal buffers?

 

[KlausST]

Sharing ports does not reduce the total power dissipation in the package. You must give more specific design limits before you can clearly choose from all the options given to you. I think a conservative max Tjcn at 150 to 200'C rise /W will be a gating item along with total costs.

Driving 3 x 420mA using 3 ports , the voltage drop at each port is 0.8V, gives a total of 3 * 0.8V * 0.42A = 1.01W
Driving 3 x 420ma using 6 ports (two shared each) the voltage drop becomes 0.2V and thus gives a total of 6 * 0.2V * 0.21A = 0.5W

So in my eyes sharing ports reduced the total power dissipation to 50%. Where am I wrong?

Klaus

 

[D.A.(Tony)Stewart]

Driving 3 x 420mA using 3 ports , the voltage drop at each port is 0.8V, gives a total of 3 * 0.8V * 0.42A = 1.01W
Driving 3 x 420ma using 6 ports (two shared each) the voltage drop becomes 0.2V and thus gives a total of 6 * 0.2V * 0.21A = 0.5W

So in my eyes sharing ports reduced the total power dissipation to 50%. Where am I wrong?

Klaus

How did you reduce voltage drop from 0.8 to 0.2 with 2 shared switches from that datasheet? They act like fixed RdsOn @ some Temp.

 

[KlausST]

How did you reduce voltage drop from 0.8 to 0.2 with 2 shared switches from that datasheet? They act like fixed RdsOn @ some Temp.

Typo:
0.2V --> 0.4V The rest of my formula was correct.

6 * 0.4V * 0.21A = 0.5W

Yes, they act as a constant R_DS_ON. So if you feed 0.42A via two channels, then each channel drives 0.21A only resulting in a 0.4V drop.

Klaus

 

[D.A.(Tony)Stewart]

my mistake
I^|^|