Fullscreen HTML5Med Res (??MB)HTML4

Modifying the Currawong Valve Amplifier . . . is it worthwhile? By Allan Linton-Smith & Leo Simpson While the Currawong amplifier has created a great deal of interest, some readers would like to see it with improved frequency response, better output transformers, more expensive valves and so on. We have investigated a number of these possibilities and you can judge for yourself whether all or any of the modifications discussed are worthwhile. M OST READERS would regard the output transformers we used as looking physically puny compared to the much larger transformers fitted to valve amplifiers in the “olden days” and we would have to agree. So could bigger and better output transformers improve the performance? Possibly. Before we had a look at that topic we had to address a query about the lowfrequency response of the Currawong. As depicted in the graph of Fig.5 on page 38 of the November 2014 issue, the frequency response had a slight upturn at around 20Hz. Some people blamed this on the relatively small The Hashimoto HW40-5 is much larger, heavier and more expensive than the Altronics M1115 line transformer. While its frequency response is flatter above 3W, the M1115 actually provides substantially lower distortion over most of its frequency range. This is likely due to its use of grain-orientated steel in the core 74  Silicon Chip 100µF capacitors at the cathodes of the 6L6 output valves. These supposedly did not allow sufficient decoupling at the lowest frequencies and the gain climbed slightly as a result. We did not agree with this contention for the following reason: increasing the cathode bypass capacitors will actually increase the low frequency open-loop gain but the effect of negative feedback will be to negate this anyway, and it will therefore have negligible effect. Thus, we ran the frequency response test with an 8-ohm load again and compared the response with 100µF and 200µF capacitors (ie, with another 100µF in parallel) bypassing the 330Ω cathode resistors. Fig.1 shows the results and as expected, there is negligible difference in the two curves. By the way, these curves are even flatter than those originally published in the November 2014 issue and we can only put this down to a slightly different valve line-up and wiring layout in the final prototype of the amplifier. We should also point out siliconchip.com.au +3 Currawong Frequency Response (revised) 28/01/15 14:38:24 11/10/14 21:43:18 M1115 vs Hasimoto Power Response 20 Load: 8Ω, analyser bandwidth: 20Hz-80kHz +2.5 +2 10 Hashimoto 7W 7 6 +1 100μF +0.5 220μF +0.0 -0.5 Power (Watts) Amplitude Variation (dBr) +1.5 M1115 7W 5 4 M1115 4W 3 M1115 3W M1115 2W 2 -1 -1.5 M1115 1W 1 -2 0.7 0.6 -2.5 -3 10 20 50 100 200 500 1k 2k 5k 10k 20k 0.5 50k 100k 20 50 100 200 Frequency (Hz) 500 1k 2k 5k 10k 20k Frequency (Hz) Fig.1: the Currawong frequency response as designed (blue) and with extra output stage cathode resistor bypass capacitance (red). Fig.2: a comparison of the power response of the M1115 and Hashimoto transformers in the Currawong at various power levels. 30/01/15 14:52:21 Altronics M1115 Frequency Response +10 30/01/15 15:02:23 Altronics M1115 THD+N vs Frequency 100 Load: 660Ω, analyser bandwidth: 20Hz-80kHz Load: 660Ω, analyser bandwidth: <10Hz-500kHz 50 +8 20 Total Harmonic Distortion + Noise (%) Amplitude Variation (dBr) +6 +4 +2 +0 1W 7W -2 15W -4 10 5 2 1 7W 15W 0.5 0.2 1W 0.1 -6 0.05 -8 0.02 -10 10 20 50 100 200 500 1k 2k 5k 10k 20k 50k 100k 0.01 20 50 100 Fig.3: frequency response of the M1115 transformer operated open loop into a 660Ω resistive load. The load resistance gives 15W at its design output voltage of 100V. that, as in any typical high-performance valve amplifier, the Currawong needs to run for at least half an hour before it produces the best performance. Now to the question of the output transformer. A number of readers have pointed out that we should have published power response curves for the Currawong as these would soon throw up the deficiencies of the Altronics line transformer. Hence we have prepared a series of power response curves and compared these to a highly regarded substitute transformer, the Hashimoto HW-40-5, made by Hashimoto Electric Ltd in Tokyo, Japan (available at more than siliconchip.com.au 200 500 1k 2k 5k 10k 20k Frequency (Hz) Frequency (Hz) Fig.4: distortion of the M1115 transformer with the same set-up as in Fig.3. The distortion is quite low at 1W but increases at higher power levels and lower frequencies. US$700 for a pair). The frequency response claimed by the manufacturer is flat from 10Hz-60kHz ±0.1dB and it has an input impedance matched specifically for 6L6 valves of 5kΩ and output taps at 4Ω, 8Ω and 16Ω. It is suitable for amplifier powers up to 40W. These transformers weigh 2.4kg each and are far too big and heavy to be mounted on the Currawong PCB, so they were externally mounted with longer leads. Fig.2 shows a number of power response curves run with the Altronics transformer and one with the Hashimoto transformer at an output power of 7W into an 8-ohm load. Looking at the curves, the Altronics transformer does lack bass power at higher levels but is quite adequate up to about 3W RMS whereas the Hashimoto transformer has a flat power response down to below 20Hz. The Hashimoto transformer was also tested for frequency response at various power levels up to 20W without negative feedback. Under this condition, the Hashimoto performs much better than the Altronics unit, as would be expected. Given that result, you might expect that the Hashimoto would produce significantly less harmonic distortion when feedback is applied (as in the normal Currawong March 2015  75 Fig.5: distortion from the Currawong with M1115 output transformers driving an 8Ω load at 20Hz & 1W. The residual is largely third harmonic and while the waveform distortion is clearly visible, it’s still somewhat sinusoidal. configuration) but surprise, surprise, it turned out that the THD+N at 1W was higher than the cheaper transformer, as shown in Fig.7. The negative feedback in the Currawong circuit is quite high for a valve amplifier and this will linearise the response and reduce harmonic distortion in the smaller transformer. Hence, the negative feedback was reduced to zero to see if the Hashimoto could do with less and therefore produce more power. It did and the best we could squeeze out of it was 20W but the harmonic distortion was a whopping 20% at 1kHz (with zero feedback). Subjective listening tests Subjective listening tests proved that the Hashimoto is a very good transformer but at more than 40 times the price of the Altronics M1115, it really is only marginally better. Of course, both transformers could deliver more power if the Currawong amplifier was run with much higher power supply rails and the circuit bias modified to suit. However, the cheaper transformer would still be deficient in power response at the low frequency end, simply because its core is not big enough. To illustrate just how good (or bad, depending on your viewpoint), we decided to do a number of tests on the Altronics M1115 transformer when driven by a high-quality solid-state amplifier. In this case, the amplifier was connected to the primary winding and the transformer was used in step76  Silicon Chip Fig.6: same as for Fig.6 but at 4W. It certainly doesn’t look like a sinewave any more! The global feedback is applying maximum bias to try to correct the waveform but the transformer is saturated and it simply isn’t possible. up mode, as a 100V line transformer. The secondary winding was loaded with a 660Ω 15W resistor (three 220Ω 5W resistors in series). In this mode, the transformer delivers 15W. Fig.3 shows its frequency response at power levels of 1W, 7W & 15W. As can be seen, it’s pretty good at 1W and obviously somewhat deficient at the low-frequency end when driven at 7W or 15W. This is due to core saturation. The equivalent THD+N curves in Fig.4 reinforce the story and you can see that harmonic distortion rises drastically at the lower frequencies and particularly at high power levels. To further demonstrate how transformer core saturation affects the low-frequency response, have a look at the scope grabs of Fig.5 & Fig.6. Fig.5 shows a 20Hz signal at 1W with the upper (yellow) trace being the transformer output while the lower (green) trace is the harmonic distortion; predominantly third harmonic at 60Hz. Fig.6 is significantly worse with a 20Hz signal at 7W. Here the output of the transformer is running well into saturation and the harmonic distortion waveform is quite a bit worse, with more higher-order harmonics. At higher power levels, the story is similar with the distortion climbing to over 60%, as can seen from Fig.4. Now let’s consider the low-frequency power response and harmonic distortion of the Currawong amplifier. This demonstrates the miracle of negative feedback. Without negative feedback applied in the Currawong amplifier circuit, the performance is pretty awful and even with the Hashimoto transformer, it is pretty ordinary. Negative feedback makes all the difference in the Currawong, as it does in any other high-performance valve amplifier. Next time you read how valve amplifiers can sound good without negative feedback, you will know that the writers are simply ignorant! Various valves A quick search of the internet will glean a lot of information, opinions and prices for various valve brands, ages and types. You will also see how many valve aficionados prefer “NOS” valves (New Old Stock) which have been manufactured up to 50 years ago but have never been used (and sometimes in the original box). If you go to www. tubedepot.com you will find more than 30 different types of 12AX7 priced from US$11.95 for a basic ElectroHarmonix right up to US$540.95 for a “Black Sable Mullard”. You may well wonder how much improvement you might get from the higher-priced valves. We would advise extreme caution. NOS valves can command high prices but it is very much a case of “buyer beware”. Such valves may have been used (definitely not “new”!) and there are even forgeries of the most popular types. If you have built the Currawong and then start swapping valves you may notice differences between similarly priced valves such as siliconchip.com.au 100 M1115 vs Hasimoto THD+N vs Frequency 11/10/14 22:04:00 Load: 8Ω, analyser bandwidth: 20Hz-80kHz 50 Total Harmonic Distortion + Noise (%) 20 10 5 2 Hashimoto 1 0.5 0.2 M1115 0.1 0.05 0.02 0.01 20 50 100 200 500 1k 2k 5k 10k 20k Fig.7: a comparison of the distortion performance of the M1115 and Hashimoto transformers at 1W without negative feedback. Surprisingly, the M1115 has much lower distortion. Fig.8: spectral response for the Currawong at 5W into an 8Ω load using the Jaycar-supplied Sovtek 12AX7 valves. The result is slightly different to that achieved when substituting valves from other manufacturers. Fig.9: spectral response for the Currawong under the same conditions as Fig.8 but with the Electro-Harmonix 12AX7 valves supplied by Altronics instead. Fig.10: spectral response for the Currawong at 1W using 6CG7 valves but with feedback enabled. Note that these valves require a 6.3V filament supply. Frequency (Hz) Electro-Harmonix (from Altronics) versus Sovtek (from Jaycar). But while these differences may be discernible and you might like one or the other depending on the type of music you prefer, objective tests will show that frequency response and total harmonic distortion are quite similar. With that in mind, you might discount subjective differences. But it turns out that the differences are real and hence perceptible, which is backed up by the different spectra for these valves. You can see the results in Fig.8 & Fig.9. In both cases, the input signal is a 1kHz sinewave and spectra show the amplitudes of the various harmonics. Apart from the multiple different brands of 12AX7 and 6L6 valves, siliconchip.com.au you could also try 6CG7s in place of the 12AX7s but then you will need to run the filaments at 6.3VAC, not 12VAC. The 6CG7 is a very linear valve previously used in TVs for vertical oscillators to maintain a nondistorted picture. These valves are now available at Altronics. The spectrum for the 6CG7 is shown in Fig.10. Note, though, that this was plotted at a 1W power level and with feedback enabled, in contrast to Figs.8 & 9. So use caution when comparing these results. There is also the possibility for using KT66 valves in place of the 6L6s. These are significantly bigger and bulkier which does look more impressive. The performance is again very similar but they are more expensive. The Currawong PCB is designed to accommodate them. Conclusion We h o p e t h a t r e a d e r s n o w understand that the Altronics M1115 transformer really does deliver quite a respectable performance in the Currawong and especially so, given its low price. Yes, we could have selected much more expensive transformers but the major increase in cost would simply not be justified in view of the small difference in performance. However, swapping valves to find which ones you prefer can be worthwhile and a lot of fun. SC March 2015  77