- Joined
- Oct 10, 2020
- Messages
- 893
- Likes
- 2,879
Almost a year ago I posted this short IMD comparison of various transformers in a iso box review by @amirm, and promised that I'll try to follow up with a more detailed post. This will be it. 
A few different types of devices will be measured and compared, here's a list of units under test:
Here's a list of tools used to perform the measurements:
Let's start with a comparison of frequency response:
As we can see, the Orchid Classic DI has the flattest response and the widest bandwidth between the tested units.
The Behringer ADI21, when footswitch is engaged performs FR shaping (as it is intended for use with acoustic guitars).
Next, let's compare the distortion profile vs frequency of all tested units (Note: max allowed input level is not the same for all units so I had to vary the level between the tests):
As probably expected, we see that the ART Tube MP exhibits the most distortion (dominated by the 2nd harmonic), while the rest of the boxes have relatively low distortion levels. However all of them perform worse than the HiZ inputs of many studio interfaces, e.g. here's how my RME Babyface HiZ input measures:
I measured also maximum input at 1% THD for all units at 40Hz and 1kHz - here are the results:
As we can see, there not a lot of frequency dependence of max input level with any of the units. Note that some of the units could take higher input levels if you engage the pad switches / change input trim levels.
Active DI boxes
Again let's start with a comparison of frequency responses:
Both look pretty good.
Let's see the distortion profiles vs frequency at the same (8,6dBu) input level:
What we can see above is that the two transformers have slightly different voltage gain (Jensen JT-DB-E is around -21,7 dB, while the stock transformer is around -16,2 dB), and we can also see that the Jensen transformer performs much better with regard to distortion in general - but especially at very low frequencies.
I tried also to measure the maximum input at 40Hz and 1kHz but I couldn't reach 1% THD at the levels I could generate easily with devices I had on hand (around 15dBu max). This means that both of these devices can take a lot of input without saturating severely.
Passive 'Reamp' / reverse DI boxes
Comparison of frequency responses:
All devices are very good in the audible band, but it is interesting to note how the Vigortronix VTX-101-1604 low frequency roll-off increases in frequency as level goes up. This is an indication of saturation, which will also be visible in other measurements.
Let's see the distortion profiles vs frequency now:
We can see how both the Jensen and Edcor transformers significantly outperform the Vigotronix one at low frequencies - even though they are driven by much higher input levels.
Let's compare how the Jensen and Vigortronix behave as we increase the input level:
As we can see, while the Jensen transformer behaves pretty consistently with increasing input level, the Vigortronix transformer saturates at low frequencies - and the frequency at which the 1% THD is reached increases together with input level.
Again, with most of these boxes I could not reliably measure the maximum input level at 40Hz and 1kHz because I could not easily generate enough clean signal level with the devices I had on hand (around 15dBu).
The exemption is the Vigortronix transformer which reaches 1% THD at 40Hz with only 0,8dBu input - at 1kHz again I could not reach the 1% THD with 15dBu input.
The voltage gain of these reamp boxes is:
Please note however that the original voltage gain of the Jensen JT-11P-1 transformer is actually around -2,4dB; I added a resistor divider at the output of this box to reduce the level slightly as I needed this for a specific use-case.
Next let's look at another very important performance aspect of 'reamp' boxes - ground loop and noise rejection performance.
For this test I generated a ground loop in my system by connecting the Phones output of a Yamaha THR10-II guitar amplifier to the RME Babyface HiZ input, and then put the reamp box in between to test how much of the ground loop noise is removed. Each graph compares the output vs the baseline noise spectrum of the soundcard (without input - ideal case), and the ground-loop noise spectrum.
Above we can see that both the Jensen and Vigotronix transformers reject the ground loop noise completely, but not so with the Edcor transformer. I assume this might be due to insufficient shielding of the transformer component itself.
This difference in performance can unfortunately be audible under certain conditions - in my experience using the Edcor transformer to reamp sounds into a high-gain amplifier input results in more induced audible noise vs when using either the Jensen or Vigortronix units.
Isolation transformers
Frequency responses at +4dBu input:
The Palmer box has a very flat and extended frequency response, while the 'pro snake' isolation cable shows evidence of low-frequency saturation.
Distortion profiles vs frequency:
As we can see, indeed there is a significant rise in LF distortion in the 'pro snake' unit.
Let's see how both units behave at a few input levels:
We see that the Palmer unit is pretty consistent, while the 'pro snake' shows rising LF distortion levels with increasing input (similar as with the Vigortronix transformer in the previous section).
Both units have around 0dB voltage gain (i.e. unity).
As in the previous section, let me show now ground loop and noise rejection performance of these two units:
Both units perform equally well here.
Conclusion
I've been looking into the performance of audio transformers and DI boxes for many years now, since I use them for various sound production and live performance use-cases. Here's a few takeaways based on my experience so far:
[EDIT 2022-05-09] @nagster added a comprehensive set of measurements of Countryman type 85 active DI, Radial J48 active DI and Tritech DB-LL1 passive DI in posts #24, #25, #26 and #27. Thanks!
A few different types of devices will be measured and compared, here's a list of units under test:
- Active DI boxes
- Orchid Electronics Classic DI Box
- Behringer ADI2 (Note: this is actually an acoustic guitar preamp + DI)
- Behringer DI100
- ART Tube MP (Note: this is actually a tube microphone preamp + tube DI)
- Orchid Electronics Classic DI Box
- Passive DI boxes (basically a transformer in a box)
- Millenium DI-E
- Modified Millenium DI-E where the stock transformer was replaced with a Jensen JT-DB-E, and circuit modified accordingly (same look as above)
- Millenium DI-E
- Passive 'Reamp' / reverse DI boxes (basically a transformer in a box)
- DIY box with a Jensen JT-11P-1 transformer
- DIY box with an Edcor WSM10K/10K transformer
- DIY box with a Vigortronix VTX-101-1604 transformer
- DIY box with a Jensen JT-11P-1 transformer
- Isolation transformers
- pro snake BJJ 301-1 isolation transformer cable
- Palmer PLI-01
- pro snake BJJ 301-1 isolation transformer cable
Here's a list of tools used to perform the measurements:
- Topping E50 DAC as signal generator
- E1DA Cosmos ADC (B grade)
- RME Babyface soundcard, first generation - measured here
- UNI-T UT60E true RMS digital multimeter
- REW software
Let's start with a comparison of frequency response:
As we can see, the Orchid Classic DI has the flattest response and the widest bandwidth between the tested units.
The Behringer ADI21, when footswitch is engaged performs FR shaping (as it is intended for use with acoustic guitars).
Next, let's compare the distortion profile vs frequency of all tested units (Note: max allowed input level is not the same for all units so I had to vary the level between the tests):
As probably expected, we see that the ART Tube MP exhibits the most distortion (dominated by the 2nd harmonic), while the rest of the boxes have relatively low distortion levels. However all of them perform worse than the HiZ inputs of many studio interfaces, e.g. here's how my RME Babyface HiZ input measures:
I measured also maximum input at 1% THD for all units at 40Hz and 1kHz - here are the results:
| Device | Voltage gain | 1% THD @ 40Hz | 1% THD @ 1kHz |
| Orchid Classic DI | -10,9 dB ('normal/0dB' input) | 15,5 dBu | 15,5 dBu |
| Behringer DI100 | -0,4 dB (pads inactive) | 3,3 dBu | 3,5 dBu |
| ART Tube MP | 0dB (but depends on gain and level settings) | 2,5 dBu | 3,5 dBu |
| Behringer ADI21 | 5,0 dB (but depends on level settings) | 6,7 dBu | 6,7 dBu |
As we can see, there not a lot of frequency dependence of max input level with any of the units. Note that some of the units could take higher input levels if you engage the pad switches / change input trim levels.
Active DI boxes
Again let's start with a comparison of frequency responses:
Both look pretty good.
Let's see the distortion profiles vs frequency at the same (8,6dBu) input level:
What we can see above is that the two transformers have slightly different voltage gain (Jensen JT-DB-E is around -21,7 dB, while the stock transformer is around -16,2 dB), and we can also see that the Jensen transformer performs much better with regard to distortion in general - but especially at very low frequencies.
I tried also to measure the maximum input at 40Hz and 1kHz but I couldn't reach 1% THD at the levels I could generate easily with devices I had on hand (around 15dBu max). This means that both of these devices can take a lot of input without saturating severely.
Passive 'Reamp' / reverse DI boxes
Comparison of frequency responses:
All devices are very good in the audible band, but it is interesting to note how the Vigortronix VTX-101-1604 low frequency roll-off increases in frequency as level goes up. This is an indication of saturation, which will also be visible in other measurements.
Let's see the distortion profiles vs frequency now:
We can see how both the Jensen and Edcor transformers significantly outperform the Vigotronix one at low frequencies - even though they are driven by much higher input levels.
Let's compare how the Jensen and Vigortronix behave as we increase the input level:
As we can see, while the Jensen transformer behaves pretty consistently with increasing input level, the Vigortronix transformer saturates at low frequencies - and the frequency at which the 1% THD is reached increases together with input level.
Again, with most of these boxes I could not reliably measure the maximum input level at 40Hz and 1kHz because I could not easily generate enough clean signal level with the devices I had on hand (around 15dBu).
The exemption is the Vigortronix transformer which reaches 1% THD at 40Hz with only 0,8dBu input - at 1kHz again I could not reach the 1% THD with 15dBu input.
The voltage gain of these reamp boxes is:
| Device | Voltage gain |
| Jensen JT-11P-1 | -3,5 dB |
| Edcor WSM10K/10K | -4,7 dB |
| Vigortronix VTX-101-1604 | 0,0 dB |
Next let's look at another very important performance aspect of 'reamp' boxes - ground loop and noise rejection performance.
For this test I generated a ground loop in my system by connecting the Phones output of a Yamaha THR10-II guitar amplifier to the RME Babyface HiZ input, and then put the reamp box in between to test how much of the ground loop noise is removed. Each graph compares the output vs the baseline noise spectrum of the soundcard (without input - ideal case), and the ground-loop noise spectrum.
Above we can see that both the Jensen and Vigotronix transformers reject the ground loop noise completely, but not so with the Edcor transformer. I assume this might be due to insufficient shielding of the transformer component itself.
This difference in performance can unfortunately be audible under certain conditions - in my experience using the Edcor transformer to reamp sounds into a high-gain amplifier input results in more induced audible noise vs when using either the Jensen or Vigortronix units.
Isolation transformers
Frequency responses at +4dBu input:
The Palmer box has a very flat and extended frequency response, while the 'pro snake' isolation cable shows evidence of low-frequency saturation.
Distortion profiles vs frequency:
As we can see, indeed there is a significant rise in LF distortion in the 'pro snake' unit.
Let's see how both units behave at a few input levels:
We see that the Palmer unit is pretty consistent, while the 'pro snake' shows rising LF distortion levels with increasing input (similar as with the Vigortronix transformer in the previous section).
Both units have around 0dB voltage gain (i.e. unity).
As in the previous section, let me show now ground loop and noise rejection performance of these two units:
Both units perform equally well here.
Conclusion
I've been looking into the performance of audio transformers and DI boxes for many years now, since I use them for various sound production and live performance use-cases. Here's a few takeaways based on my experience so far:
- All transformers I tested show rising low-frequency distortion. In extreme cases of input overload/saturation we also see low-frequency FR deviations.
- I mainly grade audio transformers based on a) maximum input, b) LF distortion and c) noise rejection performance. Usually only very expensive ones do all three really well - e.g. Jensen units are in my experience really superb, but I also find performance of the reasonably priced Palmer unit really impressive.
- Cheaper audio transformers can still perform pretty well, but you sometimes need to be careful with input levels to keep low-frequency distortion under audible levels (e.g. see the Vigotronix unit and the 'pro snake' cable). Vigortronix transformers seem to be a solid budget choice.
- Unshielded transformers (e.g. Edcor) may not always provide sufficient noise rejection for all use-cases.
- Active units usually have more consistent distortion performance, but may add noise (and obviously require a power source )
[EDIT 2022-05-09] @nagster added a comprehensive set of measurements of Countryman type 85 active DI, Radial J48 active DI and Tritech DB-LL1 passive DI in posts #24, #25, #26 and #27. Thanks!
Last edited:
