My very first breadboard

[Cressy Snr]

After yesterday, trying the AE 3K amorphous core output transformers and hearing the effect they had on the bass performance of the amp, this morning I reconfigured the Hammond output transformers for an 8 ohm load.

Simon's Hammond 1627SE outputs are 2K5 into 8 ohms so only 500R less than the AEs. The result was pretty spectacular at the bass end, giving an extended tight bass performance and an increase in the fullness of the midrange compared to the "pretend" 5K load I had been using with the deliberate mismatch on the 4R taps. Additionally, more volume was available from the amp when set correctly for the speakers. The operating point for the PX25s has hardly moved at all, statistically insignificant I would call it. Just a tiny drop in HT due to a tad more resistance from the extra bit of secondary winding that is now in play.

Now this load is about 1K less than the minimum Osram recommended load of 3K5 but the supposed increase in distortion by using a low load does not seem to be evident to my ears anyway.

I remember Paul commenting ages ago on the futility of worrying about output transformer impedance, arguing that, within reason, it was the inductive load that was far more significant to the performance of a transformer. The correct tapping for my Fostexes is 8R. Using that tapping on an ostensibly incorrect transformer produced a far better result than trying to fake a 5K load with a 2K5 transformer and losing inductance in the process.

I had not realised what I had been losing by doing this until I visited Nick yesterday. Without the support of a pair of subs that could be turned up to compensate, Nick's OBs revealed a distinct bass lightness to the sound that I hadn't known about previously.

This could have been embarrassing in bigger company

Steve

[Paul Barker]

Actually at home it's hard to see the whole picture because our minds and ears tune gradually to believe what has evolved is normal ergo "correct".

[Cressy Snr]

A bit more news on the breadboard. I decided to pull down the voltages on the PX25s; A to give them an easier time and B because the present low-impedance output transformer prefers valves to be running on a lower voltage than a high impedance transformer would demand. Current is around 30mA.

Connecting the HT supply via the 150-0-150 400 mA winding gives around 185V at the PX25 anodes with 12V of cathode bias with a GZ34 rectifier. This has obviously limited the power available at the outputs but at the same time there is a nice sense of ease to the music and a bit more detail in the treble than when I was thrashing the meshplates at 400V into too low an impedance. To operate at this voltage level really needs proper 5K or above output transformer. 2K5 was pushing the envelope a bit.


More interesting is the prospect of using the 400-0-400 winding to the output valves only, with the 150-0-150 feeding the drivers through a choke input supply. This will enable direct-coupling of the drivers to the PX25s without compromising their op points. However, to direct couple, the drivers would need changing to something like an ECC88 to enable operation at about 100V with good linearity, otherwise the voltages needed at the PX25s to maintain their linearity would be stupid.


Steve

[Cressy Snr]

Hi,
Here's a schematic for my attempt a Loftin-White cct using a half an ECC88 twin triode per channel, direct coupled to the grids of the PX25s.



The drivers use a separate supply (150-0-150 used as a 0-300) fed from a snubbed silicon bridge rectified cap input supply. High HT with a 30K (15K + 15K) anode resistor for low distortion, giving 100V at the ECC88 anode, with 2V and 9mA of cathode bias. A very very simple circuit all in all.

The potential divider at the anode of the ECC88 and the associated 33uF cap and 22K resistor allow the output signal from the driver valve to float on the cathode potential of the output triode. From what I can glean from a bit of research, this is what distinguishes the Loftin-White circuit from the ordinary direct coupled arrangement. Whether it'll work is another matter.

Steve

[Nick]

The 137 should be 237 I think.

[Cressy Snr]

The 137 should be 237 I think.

I'm going to have to get some new glasses Nick

Steve

[Nick]

The 137 should be 237 I think.

I'm going to have to get some new glasses Nick

Steve

Well, given you almost picked up a B+ connection instead of the glasses when at mine, yes, I might agree

[Cressy Snr]

I finished the Loftin White conversion this evening.

I used an ECC88 as a driver but rather than use a high HT with a large resistor, I fed the driver from a full wave solid state rectifier rather than the bridge and made it choke input. The drivers were then choke loaded, as had been the 6A6s. This arrangement couldn't have been better, as it gave almost exactly 100V at the driver anode, perfect for direct-coupling and just about ideal for the ECC88 in terms of linearity.

The solid state rectification, on the driver stage power supply, whilst using valve rectification on the output stage, gives an important safety advantage with this type of amp, as the 100V positive voltage appears at the grids of the PX25s almost instantaneously, greatly reducing the problem of incorrect bias on the outputs whilst they wait for the inevitable indirect heated input valve to come up to speed. It's not a perfect solution (what is except separate heater and HT switching?) but it is a darned sight better than having a pair of expensive DHTs sitting wide-open for five to ten seconds. Granted, the instant start will somewhat shorten the life of the ECC88, but I'd rather replace a cheap ECC88 after a few thousand hours than have an expensive direct heated power triode go bang. In short then, it's a compromise worth making.

The amp powers with no drama or rectifier arcing as my Bluebell LW was prone to do unless a strong rectifier such as a Sovtek or Mullard was used

I'll comment on the sound later but sufffice it to say that soundwise, the amp is very good indeed.

More to follow soon with a few pics.

Steve