Todor,
I have a couple of things to say in response to your questions, but first I must ask: Why are you going to all this trouble on a PR99? According to your other posts here and on the Studer list, you have at least one A80 and at least one A810. The PR99, while it's a nice home machine, and I know because I have one in my home, will never be the machine that an A80 or A810 is.
OK, you say "they" told you that the butterfly heads from the A67 can be used on your PR99. Who are "they?" And why didn't they give you enough information to do it successfully?
According to the Studer Head Catalog available from the MRL site, the basic specs of the heads are pretty close. So, maybe it will work. Or not. One of the things they (and any other manufacturer in the same position) did to make the machines hit the much lower price points was to skimp on the beefiness of the electronics. It may be that the PR99 electronics just don't have the current capability to drive those heads. But the first thing to do is not to give up, it's to try to make it work.
First, your erase problem: You must be erasing the wider track that your butterfly heads are using. That means either a full track head or a two track erase head that has the wider track. The part number for the wide track erase head from the B67 is 1.116.097. The part number for a PR99 full track erase head is 1.116.089.
If you are not able to get the bias to go beyond the peak, that means you are considerably underbiased, and operating where the distortion and modulation noise are both far from optimum. The first thing you should do is adjust the bias oscillator to the new heads. This is a tuning procedure where you adjust a couple of variable inductors to get the system to resonate with the heads.This allows the full current capacity of the bias oscillator to be delivered to the record head. The manual spells out the procedure well enough. You do have the manual, don't you? If not, immediately go to the Studer ftp site and download it. Studer have very generously put all the documentation for all the tape machines on that site, including the ReVoxes. It's a great resource and we should take advantage of it with gratitude. Search on this site...the directions to the ftp site have been published here several times.
Actually you should do that when the final erase head is in the machine, since that's part of the circuit too.
If you can get the bias set correctly, then you can look at the record frequency response. It will almost certainly have changed from what you reported above. You have EQ adjustments on the record board, so with luck they will have enough range of adjustment to achieve flat response, within the machine's specifications.
It is important to do the various steps of an alignment in the correct order, since so many of them build upon earlier steps. If you just follow the steps in the order you find them in the ReVox manual, you'll be OK.