LM3886 Power Supply Ripple and using a DMM on AC to measure Ripple

I am building a gainclone amp based on the xy pcb's and today was testing the power supply with a pure resistor load.

The power supply is this configuration :-


Transformer is 225VA 18-0-18-0
Single Bridge Rectifier
2 x 22,000uF then 2x 10,000uF caps (33,000uF per rail)
0.1uF bypass caps soldered directly on each big cap and a mylar 0.15uF bypass at psu output terminals.
centre rail of the psu and the 0-18 centre tx taps joined and go back to a star earth on the chassis which is connected to the mains iec earth.

I put a load of 800R on each rail just to test and get +-28v
I then used a cheap Uni-t DMM on AC to see if I could measure ripple and got a very odd reading of 60V AC on each rail (Not possible with 18-0-18 in). I thought I might have a bridge fault at first so I got out my scope and used that and it showed 27mV of AC ripple on each rail sitting on 28V DC


Can you use a basic DMM set to AC to measure ripple at the DC output and why would I get such an odd reading like 60V AC across each rail. Infact thinking about it, it is half mains voltage across the + and - even though the scope shows correct +- 28V DC with some ripple. I just don't get what it was reading ? Maybe a analogue meter would have been better ?

What is an acceptable amount of ripple for the LM3886TF and is it possible to reduce it without regulators ? I know the resistor I used will not be the i/p supply resistance of the 3886 circuit so this maybe a dumb test at this stage since a higher resistor should give less ripple. I guess I need to test on its correct load but have no idea what resistor I should have used just to psu test.

Sorry if the first question seems dumb but I can't get my head round that.
Last edited:
I used a digital volt meter with the AC voltage selection to read the power supply ripples. This simple test will tell you how good it is your filtering. The reading should be almost close as possible to zero. My LM3886 rails voltage read about 2mv. You should use this test just in case you do not have an scope on hand. Since you have an scope then use it to read the rails voltage. To test the meter in AC just read a battery voltage and it will read zero. Now the question it is where did you read it when you read the voltage with the meter. You should read it at the capacitors terminals and ground.
Your DMM cann't give a correct reading because of the DC voltage of the rail. Try to insert a capacitor (say, 4.7 uF film) in series with the DMM input, to separate AC component of the rail voltage. But you can as well use your scope to measure ripple... perhaps it will be more precise !

OK, that makes sense. I maybe thought the ac input of the dmm was cap coupled inside. I am sure I did this in the past using a more expensive meter but this is a 10 dollar dmm I used. The scope made more sense. Maybe only works on some meters if cc.

I'll test the ripple properly when wired to the real PCB. The higher load resistance should give low ripple looking at the formulae.
I have four DMM, three fairly cheap (one is very old) hand held battery powered and one mains powered bench style and rms reading.

All give a good reading of Vac when reading the DC supply.

I have compared to the AC coupled input of a scope and can see that Vpp of the DC ripple from the scope is very close to 3*Vac read from the DMM.
If you don't have a scope, then use the capacitor coupling to your DMM and take a reading and then use the direct coupling to your DMM and compare the readings for a variety of test arrangements/amplifiers. Build up some experience of how your DMM measures AC ripple when in the presence of high DC offsets.

I also tried testing my school's £3 DMM on this DC+AC test and they too performed well on their 199.9mVac scale.
Last edited:
To be correct, a true rms meter must take into account both AC and DC because what it gives you is the heating value of the voltage if it were applied to a resistor. Almost all meters are AC coupled on AC, and don't do this, with the exception of a few that have a switch or software setting to put them in a DC coupled mode. As said above, you need to AC couple the meter for ripple readings to get the right result if it's giving you some odd number.