Is it possible to specify a model with correlated random effects and different residual variance for 2 conditions?

Hello,

I'm still a little stuck with specifying the model I want.

Participants were tested under 2 Delivery conditions (B and M) at a  
variety of Speeds.
I'm interested in how dv varies with speed and delivery type and in  
individual differences.
I expect greater change in dv with speed for the B condition and   
greater subject to subject variability in the B condition than the M  
condition.

I have tried some simple models but don't understand how to specify  
what I want: A model that has correlated random effects AND different  
residual variance for each level of delivery (or is that implicit in  
model 1)?

model 1: correlated random effects
fm1 = lmer(dv ~ Speed*Delivery + (Speed | Participant))

model 2: uncorrelated random effects
fm2 = lmer(dv ~ Speed*Delivery + (1 | Participant) + (0 + Speed |  
Participant))

model 3: same as model 2 but with different residual variance for each  
level of delivery (where Bind and Mind are indices of B and M -  
similar to https://stat.ethz.ch/pipermail/r-sig-mixed-models/2007q3/000248.html)
fm3 = lmer(dv ~ Speed*Delivery + (0 + Bind | Participant) + (0 + Mind  
| Participant) + (0 + Speed | Participant))


Thank you so much,
James