One study examined the role of testosterone supplementation in hypogonadal men with ED. These men were considered nonresponders to sildenafil, and their erections were monitored by assessing nocturnal penile tumescence (NPT). After these men were given testosterone transdermally for 6 months, the number of NPTs increased, as did the maximum rigidity with sildenafil.  This study suggests that a certain level of testosterone may be necessary for PDE5 inhibitors to function properly.
In this study, ED proceeded CVD in almost 70% of cases. Similarly, many men with ED have been found to have pre-existing CVD. A study by Vlachopoulos et al evaluated the incidence of asymptomatic CVD in 50 men with ED.22 These authors found that 19% of men with ED had asymptomatic CVD. Similarly, Mulhall and colleagues found that 20% of men presenting with ED and vascular insufficiency on penile duplex had asymptomatic CVD.23
Diabetes is a well-recognized risk factor for ED. A systematic review and meta-analysis found that the prevalence of ED was 37.5% in type 1 diabetes, 66.3% in type 2 diabetes, and 52.5% in diabetes overall—a rate approximately 3.5 times higher than that in controls.  The etiology of ED in diabetic men probably involves both vascular and neurogenic mechanisms. Evidence indicates that establishing good glycemic control can minimize this risk.
You have to desensitize yourself to stimulation. One way to do this is to masturbate about an hour beforehand, achieving another orgasm shortly after that will be more difficult. If you continue to have trouble, talk to your partner about changes to the timing of your sexual routine, you may need to give them oral or manual stimulation first or after you ejaculate, etc.
It appears that testosterone has NOS-independent pathways as well. In one study, castrated rats were implanted with testosterone pellets and then divided into a group that received an NOS inhibitor (L-nitro-L-arginine methyl ester [L-NAME]) and a control group that received no enzyme.  The castrated rats that were given testosterone pellets and L-NAME still had partial erections, a result suggesting the presence of a pathway independent of NOS activity.