monarch vu meter - not reading 0VU correctly

I have ten Monarch VU-100C panel meters that a friend gave to me a while back. I'd like to use them for mic preamps at some point. When I hook them up to a 100Hz +4dBu signal, they all read about 2dB low. They already have the requisite 3.6k resistor installed internally.

Inside there is a small piece with four leads and a set screw. I'm guessing this contains the rectifiers. The set screw only turns about a quarter turn. It causes the meter to read -5dB instead of -2dB. There is no middle ground.

Can I just change the 3.6k resistor to something lower to make it read correctly at 0VU? I'm not too concerned about precise ballistics. I just find it odd that all ten meters read low. They may just be cheap junk and not worth my time.

http://tacticalworks.us/img/meter1.jpg
http://tacticalworks.us/img/meter2.jpg
http://tacticalworks.us/img/meter3.jpg

Note: I set the zero adjustment before I ran a signal through it. It was off in the first pic.

That is a selenium stack rectifier and the screw is probably not an adjustment, but there just to hold the stack together. I personally have never seen an adjustable rectifier with a "fine tunning" screw.
Do you have a different "known good" VU meter you can use to check that your 1.23 volts actually produces 0 VU?
What do they read at 1 KHz? They don't look like that high a quality meter, like a Weston or API, so you may just be seeing the effects of "you get what you pay for".

The other mechanical vu meters I have are in my sony console and otari 8-track. Nothing I could easily disconnect. I did however sweep a 20 to 20000 Hz signal through the monarch, and it tracked just fine over the whole range.

I'm probably not going to bother with them since they were so cheap (free). I just had them sitting around so I thought I'd give them a shot.

EDIT: I used 100 Hz for the pic beacuse I know that my Fluke is accurate in that range. I have tested it on other sources in the past and found it to be spot on.

Well, you could always drive the meters with a buffer, which is the prefered way and give it a couple db of gain to play with. That would be one solution.