Blaupunkt Melbourne 120 vs voltage dip during engine start

I replaced factory-mounted car radio with new fancy Blaupunkt (mostly because of possibility to playback mp3). The radio works fine but I noticed that every time I start-up the engine the radio switches off because of supply voltage dip (engine starter drains lots of current). I try to do ecodriving which includes switching engine off when I expect that I will wait for possibility do drive further more than 1 minute and in such cases radio boots up like another 50...60 seconds of my valuable time. What's funny - there were no such problem with factory-mounted CD30.

I took Melbourne 120 from my, connected to regulated power supply and measured that the Blaupunkt product goes off when voltage falls below 9 Volt. On this picture it still works but it was enough to go 100mV lower to cause reset.

I didn't find anything trustful about car battery voltage during engine start up so I did real-life measurements with my 1.2 liter Opel Corsa. This is battery voltage measured during starting engine. You can notice that voltage falls to about 7 Volts, after around 100ms voltage reaches 9 Volts and than goes rather up.

Avoiding this voltage dip would keep the radio alive during engine start. I came up with such an circuit idea:

There are two ways the radio can be supplied: from the car battery through D1 or from the buffering capacitor C1 through D3. D1 and D3 works so the Blaupunkt will be supplied from the voltage which is momentary higher. During almost all time Melbourne will be supplied through D1 from car battery. C1 will be charged thrugh D2 to voltage which is almost equal to car battery. During engine starting battery voltage will fall down but thanks do D2 C1 will not be discharged. When car battery voltage is lower than C1 + 1.4 Volt the radio will be supplied from C1. Of course it will take a moment to discharge C1 but this is only matter of 100ms for battery voltage to go above 9 Volt again. 

Now, let's calculate C1 value to find out if it such solution is possible. From definition of capacitance:

C=Q/U

Definition of current is:

I=Q/t 

which can be transformed to

Q=tI

lets put Q intu definition of capacitance and we get:

C=tI/U

Where:

C is capacitance

t is discharge time

U is voltage drop across capacitor.

Let's calculate what voltage drop do we accept: before engine startup voltage across battery is around 12.4 Volt, on C1 voltage will be 0.7 Volt lower due to D2 so 11.7V. To keep Melbourne alive voltage across C1 must be higher than 9 Volt + 0.7 Volt so C1 voltage at the end of discharge is 9.7 Volt. 11.7 - 9.7 Volt = 2V.

Time read from oscillogram above is around 100ms, after 100ms voltage on car battery will raise and will be high enough to supply the radio. Let's assume that during this time the radio will be consuming 2 Amper.

C=0.1[s] 2[A] / 2[V] = 0.1[F]

So we are looking at 0.1 Farad capacitor with voltage rating of minimum 16 Volt. Conventional electrolytic capacitor would be huge so I looked for so called ultracaps aka goldcaps.

After short searching I decided to choose Su'scon 5.5F 2.5V. I had to use 6 capacitors in series to achieve voltage rating of 15V which is barely enough to install it in a car. I choose capacitors which have values 50 times bigger that my calculations say, it is not because I distrust my calculations - the only one and simple reason is that this capacitors were cheap. You may be tempted to choose 5V ultracapacitors but be aware - usually they are designed to backup RTC and some memory and they have high ESR value (what basically means that they can't provide too much current). 

Plug and play - My circuit is equipped with male and female ISO connector so it can be simply connected between the radio and original cable harness. 

Here is my prototype ready to go:

It works!

I connected it and surprisingly inrush current didn't blow a fuse. I didn't even bother to look at fuse and see what its rating, I just assumed that there has to be some fuse on radio supply line (and another in the Balupunkt itself). SO far I had no time to make exact measurements and test what does voltage look like but one is for sure - it works, I can start up the engine and Melbourne 120 works without any interruption.

One more thing has to be measured - leakage current. I tired to test it with my desktop power supply but with no success. To do such a measurement you need really stable voltage, in case of 5F each millivolt of voltage change causes flow of couple of hundreds of uA needed to charge or discharge the capacitor (depending on direction of this voltage change)